(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "The Dehiscence of Anthers by Apical Pores"

STOP 



Early Journal Content on JSTOR, Free to Anyone in the World 

This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in 
the world by JSTOR. 

Known as the Early Journal Content, this set of works include research articles, news, letters, and other 
writings published in more than 200 of the oldest leading academic journals. The works date from the 
mid-seventeenth to the early twentieth centuries. 

We encourage people to read and share the Early Journal Content openly and to tell others that this 
resource exists. People may post this content online or redistribute in any way for non-commercial 
purposes. 

Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early- 
journal-content . 



JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people 
discover, use, and build upon a wide range of content through a powerful research and teaching 
platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit 
organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please 
contact support@jstor.org. 



THE DEHISCENCE OE ANTHERS BY APICAL PORES. 

BY J. ARTHUR HARRIS. 

Introduction . 
history of the present investigation. 

Durinsr the summer of 1901, I became much interested 
in the pollination of Solarium and Cassia and, assisted by 
a student, published a paper giving the results of our inves- 
tigations. At the time of the preparation of that paper, I 
was fully aware of a number of points meriting further 
field study and especially comparison with similar forms 
in other genera, and, upon taking up my work at the 
Missouri Botanical Garden, was glad to avail myself of 
the facilities offered by the herbarium and library for a 
comparison of the floral structures of other species of 
these large genera and for the accumulation of notes on 
some analogous types of flowers in other families. 

While working on Solantim and Cassia, I was im- 
pressed by the close resemblance of the floral constitution 
of these systematically widely separated genera and by the 
similarity of their ecological relations. My interest was 
deepened when I observed that the same close agreement 
in the structure and frequently in the ecology of apically 
dehiscent forms is to be found in other families. Later, 
there came the question whether the great differentiation 
of Solarium, Cassia and the Melastomataceae in South 
America and the occurrence there of several smaller genera 
with apically dehiscent anthers might be significant. Obvi- 
ously the question could be answered only by a compara- 
tive and statistical study of all genera and species showing 
dehiscence by apical pores, and the collection of data for 

(167) 



168 MISSOURI BOTANICAL GARDEN. 

this purpose was immediately begun by examining Engler 
and Prantl's Die Natiirlichen Pflanzenfamilien, Bentham 
and Hooker's Genera Plantarum and several of the more 
important floras, for the purpose of preparing a descriptive 
list of all forms in which the anthers have been described 
as opening by apical pores. This was intended to serve 
as a foundation for structural comparisons, and so far 
as might be possible, for a comparative treatment of 
the floral ecology of these forms. 

The data secured and the conclusions which they seemed 
to indicate proved highly interesting and were presented in 
April, 1903, to the Faculty of Washington University as a 
thesis for the degree of Doctor of Philosophy. Since that 
time I have been able to secure a large amount of valuable 
material for study and have devoted much of my time 
which has been available for research to this problem. 
My manuscript has grown so large that it is inadvisable 
to publish it in full at present, especially since I have in 
hand further studies which I hope will add materially to 
its value but which cannot be completed for some months. 
It seems best to present an abstract outlining the problem 
and giving the essential points in the conclusions which 
seem justified by the comparisons so far carried out, with 
a much condensed statement of the data upon which they 
are based. The detailed treatment of the subject is being 
increased by structural, comparative, and ecological studies, 
and I hope to publish it in full, possibly with discussions 
of other floral types, at some future time. 

I wish to avail myself of this opportunity to express 
my obligation to several gentlemen who have had the 
kindness to send material from their herbaria to the Mis- 
souri Botanical Garden for my use. These are : Dr. F. 
Kurtz, Universidad Nacional, Cordoba, Argentina ; Dr. 
E. A. Philippi, Museo Nacional, Santiago de Chile; Dr. 
W. E.. Guilfoyle, Botanic and Domain Gardens, Mel- 
bourne, Australia; Dr. J. M. Wood, Durban, Natal; Dr. 



THE DEHISCENCE OF ANTHERS BY APICAL, PORES. 169 

J. C. Willis, Royal Botanic Gardens, Ceylon; Dr. M. 
Treub, Jardin Botanique, Buitenzorg; Dr. J. H. Maiden, 
Government Botanist, Sydney, N. S. W. ; Dr. I. Urban, 
Koniglicher botanischer Garten und Museum, Berlin; 
Dr. T. Durand, Jardin Botanique de l'fitat, Bruxelles ; and 
Dr. N. L. Britton, New York Botanical Garden. 

This material, which was asked for only after the most 
of the data for this paper had been gathered and classi- 
fied, has been consulted in the preparation of these pages 
but is being used mainly for a histological investigation of 
these types of anthers, and so my future memoir will owe 
even more to their generosity than the present one, which 
is very largely a library contribution. Perhaps nowhere 
else could this paper have been prepared so easily as at 
the library of the Missouri Botanical Garden, with its lai-ge 
collections of living and preserved plants also accessible for 
consultation, and I wish to express my gratitude to the 
Director, Dr. William Trelease, not only for access to 
these facilities, but for the most generous conditions 
under which I have used them while connected with 
the Garden and with Washington University. For 
several months Mr. Dean H. Rose has assisted me 
in this and in other research work, and while the most 
of his time has been given to phases of the subject which 
will be treated later, I wish to express in advance my 
hearty appreciation of the earnest and efficient service he 
has given me. My sister, Miss Nellie L. Harris, has been 
of great help to me in the preparation of the statistical 
portions of the paper. 

EARLIER INVESTIGATIONS . 

That in a comparatively limited number of Phanero- 
gams, the pollen is shed through terminal pores instead of 
longitudinal slits is a fact of general knowledge which has 
been utilized in the characterization of families, genera, and 
species. 



170 MISSOURI BOTANICAL GARDEN. 

The method of the shedding of the pollen in the Angio- 
sperms is a subject very briefly treated in the general works 
on morphology and physiology. Goebel in his Organog- 
raphie calls attention to the obscure nature of many 
points in this phenomenon, and Coulter and Chamberlain 
in their recent work on the morphology of Angiosperms 
speak of the need of a thorough investigation of the sub- 
ject, especially from the developmental point of view. 
Most general works treat this point with the same or greater 
brevity and need not be mentioned separately. 

Histologically, anthers have been investigated and their 
minute anatomy interpreted more or less satisfactorily by 
several writers, among whom may be mentioned Mohl, 
Purkinje, Schrodt, Leclerc du Sablon, Chatin, and Stein- 
brinck. 

So far as I am aware, no special treatment of the api- 
cally dehiscent anther has been attempted from a morpho- 
logical, histological or ecological point of view. 

PURPOSE OF INVESTIGATION AND SCOPE OF PRESENT PAPER. 

My purpose in undertaking the present investigation 
was threefold: 1. The compilation of a systematically 
arranged descriptive list of genera or species showing dehis- 
cence by pores, for use in future biological work. 2. The 
demonstration of any similarity of structure in apically 
dehiscent forms of different systematic affinities or the ex- 
istence of possible correlative modifications in the parts of 
the flowers showing this method of dehiscence, if such exist. 
3. The establishment or the refutation of the hypothesis 
that the flora of certain of the main divisions of the earth's 
surface is richer in apically dehiscent forms than that of 
others. 

These three purposes were quite satisfactorily accom- 
plished in the spring of 1903. During the time which has 
elapsed since "the writing of my first manuscript I have 
added practically no new genera or species to my list. I 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 171 

have, however, been able to supplement my descriptive 
data very materially by a comparison of the rich materials 
generously supplied by the directors and curators of botan- 
ical gardens and herbaria in various parts of the world. I 
have also been able to carry out several detailed and 
time-consuming comparisons between the apically and the 
longitudinally dehiscent members of the several families in 
which dehiscence by apical pores has been described. This 
I did to determine whether the type of floral structure 
which is repeated with such uniformity by the apically dehis- 
cent members of several little related families, is of frequent 
occurrence in these families, or whether the form of the 
anthers and the general habit characteristic of these apic- 
ally dehiscent genera or species are aberrant in the syste- 
matic groups to which they have been assigned by taxon- 
omists. 

I have been able to satisfy myself more fully concerning 
the distribution of forms by tabulating according to arbi- 
trarily limited fioristic regions the distribution of all the 
genera of Phanerogams. The" distributional phase of the 
problem is still far from satisfactory and must necessarily 
long remain so. These tabulations, however, confirm my 
early impression of the greater richness of the flora of 
some regions in species with apically dehiscent anthers. 

Floral structures are to be fully interpreted only through 
a knowledge of their ecological relations, past and present. 
I have been able to compile a considerable series of such 
data bearing upon the forms . under consideration and while 
these, with the few observations which I have been able to 
make, represent only a beginning, they seem to justify 
some very suggestive hypotheses for further investigation. 

Convinced that the great systematic differentiation 
attained by the Apidae in South America might have a 
significant bearing upon the problem, I have tabulated the 
distribution of this family, and byway of comparison, that 
of all families of the Hymenoptera. 



172 MISSOURI BOTANICAL, GARDEN. 

It is a pleasure to say that the work which has been 
done since the spring of 1904 has not materially modified 
the conclusions then reached, but has added more convinc- 
ing evidence from many sides. 

In this place I shall not give a systematically arranged 
list of the forms I have considered. This, with the neces- 
sary bibliographic citations and descriptive text and illus- 
trations, would require far more space than is available. 
I shall rather divide the former into groups or types and 
describe these briefly, mentioning, when desirable, the 
names of the genera assigned to the particular types, with 
their geographical distribution, and comparing their struc- 
ture briefly with the forms to which they are systemati- 
cally related. It is to be understood that these groups or 
types are not sharply defined categories, but transgres- 
sive assemblages. The justification for the division of 
the material into these groups or types is found in the fact 
that the majority of the species do fall clearly into one or 
the other of the groups, while those which represent transi- 
tions or are quite aberrant in structure are much rarer. 
These groups are to be regarded as tentative, and, while 
for the most part natural, they are established primarily for 
the convenience of the present discussion. Their great 
justification is to be found in the convenience of treat- 
ment and interpretation which does result from their use. 

Four of these types I shall consider very briefly, merely 
describing in general terms their differential characters. 
The fifth, sixth and seventh will be treated in greater 
detail, and in addition to the description and comparison of 
the morphological features, I shall discuss their ecological 
relations and geographical distribution. 

la this place I shall not review the work of others on the 
microscopic structure of the apically dehiscent anther or 
present the results of my own studies in this phase of the 
problem. I wish here to confine my attention to a discus- 
sion of the general morphological features and the ecologi- 



THE DEHISCENCE OF ANTHERS BT APICAL PORES" 173 

cal relations of three groups, leaving the detailed treatment 
of these and the other types until a later time. 

It seems hardly necessary to define here the terms dehis- 
cence by apical pores, or apically dehiscent anthers. The 
terms have been used in the same sense as that in which 
they are regularly used by systematists. It may be proper 
to remark, that a precise definition offers considerable diffi- 
culty . Some genera have anthers opening by a single 
pore, sometimes at the tip of a rostrate prolongation of the 
anther, which is never increased down the sides by longitu- 
dinal slits ; but, except for the Melastomataceae, in an 
equal or greater number of forms the pores are soon con- 
tinued more or less down the sides in longitudinal slits. 
Between dehiscence by the most perfected terminal pore or 
pores and the primitive longitudinal dehiscence there is a 
complete series of gradations. In this paper I have fol- 
lowed quite closely the usage of the best systematists in 
the description of my material. If an anther shows well 
defined openings at the tip and there is reason for believing 
that the pollen is shed largely through these pores, it is 
generally designated in descriptive works and floras as api- 
cally dehiscent, or described: •« anthers dehiscing by apical 
pores soon continued down the sides in longitudinal slits." 
Only living material would permit of an absolute deter- 
mination. A large proportion of the anthers are thinner 
along the lateral furrows where dehiscence usually occurs 
in forms in which the anthers do open by lateral slits, 
and in herbarium material they are liable to be broken 
along this line even when such might not occur normally 
in the living plant. Sections cut from anthers which have 
not broken along this line are very apt to break apart at 
this point. It is far from easy or even quite impossible to 
determine whether the lateral slit is a natural occurrence 
or whether it is due to manipulation. These points will 
be considered for the forms examined in my final paper, 



174 MISSOURI BOTANICAL GARDEN. 

but for the present I have used the term in the rather gen- 
eral sense in which it is employed in descriptive botany. 

In this paper I have omitted all reference to the literature. 
It is proper to state that I have examined all the most 
important systematic works treating of the forms here dis- 
cussed as well as the special literature of floral ecology. 
These citations I hope to include with ample descriptions and 
illustrations of all of the forms considered, in the final paper.* 
I have also sought to avoid all references to theoretical 
considerations and to confine myself as strictly as possible 
to the facts and the problem of grouping them in as logical 
and significant a manner as possible. 

The Types of Apical/ly Dehiscent Anthers. 
The classes recognized are the following : — 

1. ARACEOUS TYPE. 

2. GRAMINEOUS TYPE. 

3. POLYGALACEOUS TYPE. 

4. ERICACEOUS TYPE. 

5. DIIiLENIACEOUS TYPE. 

6. SOLANUM-CASSIA TYPE. 

7. MELASTOMATACEOUS TYPE. 

The first four are to be merely touched upon in the 
present paper. The last three, on the other hand, are to 
be considered in some detail ; they are characterized by 
many similarities of structure and doubtless of ecological 
relationship as well, and while, as it seems to me, they 
are separated, on the whole, by sharp mean differences, 
there is of ten difficulty in drawing the limits of the groups. 
Since they are to be understood only when examined com- 



* Most of the titles to which reference has been made by the author's 
name only in this essay, may be found in the bibliographies in the English 
edition of Miiller's Fertilisation of Flowers, the continuation of this list 
from 1883 to 1889 published by MacLeod in the Botanisch Jaarboek. 2 : 
195-254, 1890, or the more complete bibliography of Knuth's Handbuch 
der Bliitenbiologie. 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 175 

paratively, their essential points of difference may be 
stated. 

The Ericaceous type as limited in this paper contains 
some forms which are closely related to those of the three 
following, but they will be left out of consideration here. 

The Dilleniaceous, Solanum -Cassia and Melastomataceous 
types haVe in common a widely patent corolla, or at least 
a widely expanded corolla limb, and elongate, usually 
linear, basifixed anthers. The Dilleniaceous type differs 
from the other two in the possession of a larger number of 
stamens, with anthers inserted on generally long filaments, 
and often separate styles. The Solanum-Cassia type 
differs from the Dilleniaceous in the number of stamens, 
their much reduced filaments, and their frequent approxi- 
mation around the style. The Melastomataceous type is 
characterized by anthers of the same general form as those 
of the Solanum-Cassia type, but it is sharply separated by 
the elongate and highly differentiated filament. 

I. ARACEOTJS TYPE. 

Flowers usually densely crowded on a spadix ; perianth 
generally much reduced ; filaments usually shortened ; 
anthers free or very often connate in a synandrium, biloc- 
ular or multilocellate, each locule dehiscing by a more or 
less irregular apical rent, or in a few cases, through a pro- 
duced tubular process ; pollen free or conglomerate in a 
vermiform column. 

To this type belong many of the genera of the Araceae 
and Balanophoraceae. Other genera of these families shed 
their pollen through longitudinal slits in the usual way. 
The Loranthaceae and Lacistemaceae furnish interesting 
forms for comparison. 

The geographical distribution and the floral ecology of 
this type will not be discussed here. It is clearly distinct 
from others to be treated later and to which I wish to 
direct especial attention. 



176 MISSOURI BOTANICAL. GARDEN. 

II. GRAMINEOUS TYPE. 

In the characteristic anthers of the grasses the locules 
usually open by longitudinal slits extending from above 
downward, but more rarely (in some Andropogoneae ) by 
a hole at the apex, often finally continued down the side in 
slits. This type likewise is clearly distinct from all 
others, and I shall not consider it further in this place. 

III. POLYGALACEOUS TYPE. 

Flowers strongly zygomorphic, with highly differentiated 
perianth ; androecium included in a ventral carina, usually 
monadelphous ; anthers short, generally cupuliform, basi- 
fixed, dehiscing by a large, upper, terminal opening em- 
bracing both locules (Polygalaceae) or by two terminal 
pores (Fissicalyx^) . 

This group contains the Polygalaceae and the single 
genus of the Papilionatae in which the anthers open by 
apical pores. In structure and geographical distribution, 
this is an interesting type, but cannot be treated in detail. 

IV. ERICACEOUS TYPE. 

Under this type I have grouped with its special represen- 
tatives a number of structurally very dissimilar forms. 
This has been done largely because of the systematic rela- 
tionship of the material ; it has been deemed best to treat 
all the Ericaceae together and in this paper I have placed 
with them a few other genera systematically related or 
similar in their floral structure. This type will not improb- 
ably be broken up later, but at the present time I do not 
care to treat in detail the various forms which I have here 
assigned to this type, and since they are, for the most part, 
quite distinct from those to which I intend devoting the 
body of this paper, they may be passed over with merely 
a general description. 

Flowers hermaphrodite, regular or rarely zygomorphic; 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 177 

■corolla gainopetalous, globose, urceolate, tubular, infundi- 
buliform, campanulate or almost rotate, generally 4—5 -lobed, 
very rarely divided to the base ; stamens usually twice as 
many as the lobes of the corolla, equal or alternately un- 
equal, free or nearly free from the corolla, exserted or 
more generally included ; filaments usually long, various in 
form ; anthers basifixed or dorsifixed at the base or below 
the middle, rarely at the apex, dehiscing by large apical 
pores or more or less extensive slits, sometimes produced 
in one or two often very long terminal tubes opening at 
the top with small- pores or more or less extensive slits, 
sometimes provided dorsally with aristae or setaceous 
appendages. 

All Ericaceae are here included iu this type. Qalanthus 
and Leucojxim of the Amaryllidaceae, with their pendulous 
flowers and perianth closely approximated around the an- 
droecium, may also be tentatively assigned to this type. 
JPen(ap7t7/lax, the type of a family from China, may best 
be treated here, as may also Clethra of the Clethraceae, 
Oostaea of the Cyrillaceae, and Ghimaphila and Pirola of 
the Pirolaceae. 

The floral ecology of this group is not discussed here. 
Our knowledge is chiefly of the Ericaceae and the two 
genera from the Amaryllidaceae. Many of the data bearing 
upon these will be found recorded in the Handbuch of 
Knuth. 

V. DILLENIACEOTJS TYPE. 

Both whorls of perianth usually developed, but one or 
both sometimes reduced, usually campanulate or rotate in 
disposition; stamens indefinite in number; filaments long 
or short, free or variously united ; anthers mostly elongate, 
basifixed; flowers usually actinomorphic throughout, but 
androecium sometimes zygomorphic, gynoecium of distinct 
or variously united carpels ; flowers generally highly col- 
ored and conspicuous. 

12 



178 MISSOURI BOTANICAL GARDEN. 

The sharp distinction of several of the members of this 
type from, those of the following classes is quite impossible, 
but it seems helpful to make the division. 

This class exhibits, with the possible exception of the 
Ericaceous type, the widest range of form and structure of 
any recognized. To it I have assigned four species of 
Hibbertia, jLcrolrema Schumacheria, Dillenia and Sau- 
rauia of the Dilleniaceae, Elaeocarpxis, Sloanea, Vallea, 
Axistotelia and Dubouzetia of the Elaeocarpaceae, Ochna, 
Lophira and Cespedesia of the Ochnaceae, Tremaniheva of 
the Theaceae, Bixa of the Bixaceae, and Plagiopleron of 
the Flacourtiaceae. 

Dilleniaceae. 

The flowers of the Dilleniaceae are hermaphrodite, polyg- 
amous or dioecious, with usually five persistent sepals 
and as many deciduous petals which are patent during 
anthesis. The stamens are indefinite, usually numerous, 
and various in form, the gynoecium is usually composed 
of several more or less united carpels with free and for 
the most part widely divergent pistils with simple stigmas. 

In Hibbertia (about 100 species, mostly confined to 
Australia) the dehiscence is said to be by pores in one sec- 
tion represented by 4 species, but I am not sufficiently 
acquainted with the forms. A.crotrema (about 12 species 
of India, Farther India and Ceylon) has many — 15—50 — 
stamens, free or more or less united into 3 or 4 bundles, 
and linear or sometimes ovate anthers generally opening 
by two apical pores. Schumacheria (with 3 species in 
Ceylon) is distinguished by zygomorphy. The linear or 
broadly linear anthers open, according to Gilg, at the top 
with two little holes which gradually increase more or less 
down the sides in longitudinal slits. Dillenia (of about 
23 species distributed over tropical Asia, the Indo- Malay 
region, New Guinea, the Philippines, and Australia) has 
usually very long anthers which generally open at the top 
by two pores which may sometimes become confluent into 



THE DEHISCENCE OF ANTHERS BY APICAL POEES. 179 

one or may be increased down the sides in longitudinal 
slits. The flowers are often very large and conspicuous; 
Saurauia (with about 98 species of tropical Asia and 
America, rare in Brazil and Guiana) has the anthers turned 
outward in the bud, but reversed and erect at the time of 
flowering, each of the two short, sometimes somewhat 
divergent locules opening by large pores or, more rarely, 
by longitudinal slits. 

An indefinite number of stamens is characteristic of the 
Dilleniaceae, but in many forms a marked tendency to 
numerical reduction is observable, as is well seen in Hibber- 
tia, but from the data available I am unable to draw any 
definite conclusions as to the relation between the number 
of elements in the androecium and the method of dehis- 
cence. The forms of filament and anther are quite 
various. Apical dehiscence is confined almost exclusively 
to the elongate, basifixed anther. With the exception of 
one genus, the apically dehiscent forms all occur in the 
Dillenioideae with basifixed anthers and, in the exception, 
Saurauia, the versatile anthers are attached near one end 
and assume a more or less erect position at the time of 
flowering. 

Elaeocarpaceae . 

An important difference between the Elaeocarpaceae and 
the preceding family is found in the gynoecium. In the 
Elaeocarpaceae, the pistil is simple and filiform with only 
a possible indication of the compound nature of the ovary 
in the slightly lobed stigma of some forms. 

JElaeocarpus (over 100 species of tropical Africa, Asia, 
Australia, the Pacific Islands, New Caledonia and New 
Zealand), Sloanea (50 species in the tropics of both hem- 
ispheres), Vallea (3 species in the mountains of New 
Granada and Peru) and Aristotelia (7 species, 3 in Australia, 
3 in New Zealand, 1 in Chile) have campanulate or more or 
less patent, often highly colored and fringed perianth seg- 
ments ; stamens indefinite in number ; anthers linear, basi- 



180 MISSOURI BOTANICAL GARDEN. 

fixed, obtuse or provided with a terminal appendage, 
dehiscing by 1 or 2 apical pores, sometimes continued 
down the side in slits. 

In Crinodendron (2 species in Chile) and Anthdloma 
(2 species in New Caledonia) the corolla is urceolate 
and the anthers open by longitudinal slits, which may open 
more widely above or originate in a poriform opening, or 
may be confined to the upper portion of the anther. JDu- 
bouzetia (represented by 1 or perhaps more species in New 
Caledonia) has in some ways a very similar floral structure, 
but the dehiscence is by a single bilabiate terminal pore. 

The Elaeocarpaceae is a highly interesting group and I 
am far from satisfied with the treatment which can be 
given it in this paper. 

Ochnaceae. 

All genera of the Ochnaceae are to be considered in a 
paper on apically dehiscent anthers. The flowers show a 
considerable diversity of form and are to be referred to 
different types. 

Ochna (about 25—30 species distributed over tropical 
Asia and Africa, only a few species — 3, fide Flora Capen- 
sis — in the Cape region) has numerous stamens with long, 
filiform filaments and oblong or linear, basifixed anthers 
dehiscing by apical pores or longitudinal slits. It belongs 
clearly to the Dilleniaceous type. Lophira (1 species in 
central and west Africa) with the linear anthers dehiscing 
by short, terminal, almost poriform slits, belongs here, as 
does also Oespedesia (3—4 species in Peru, New Granada 
and Panama) with 40—60 elongate, linear, curved anthers 
on short filaments, all turned toward the same side of the 
flower during anthesis. 

Elvasia and Oodoya may be mentioned here as forms in 
which the stamens are more numerous than is common in 
the Solanum-Cassia type, but for other structural reasons, 
these, as well as the other members of this family, will be 
considered with the members of that group. 



THE DEHISCENCE OF ANTHEKS BY APICAL POKES. 181 

Theaceae. 
With Tremanthera (1 species in New Guinea), the single 
apically dehiscent genus of the Theaceae, I am not suffi- 
ciently acquainted, but the indefinite number of stamens 
with long-ovate anthers would indicate that it should be 
placed in this class. 

Bixaceae. 

In the Bixaceae, JBixa (1 species of tropical America, 
now widely distributed in the tropics) is a unique form 
characterized by its numerous, horse-shoe shaped anthers 
on long filaments. It may most conveniently be placed in 
this type. The other two apically dehiscent genera will be 
treated under the Melastomataceous type. 

VI. SOIANUM-CASSIA TYPE. 

Perianth usually quite large, mostly actinomorphic, seg- 
ments campanulate or more generally patent or reflexed in 
disposition ; androecium of few members, usually 5 or 10, 
very rarely as many as 15, staminodia sometimes present 
as reduced members of these numbers in zygomorphic 
forms or more rarely from a multi-staminate androecium ; 
filaments much reduced in length ; anthers basifixed, oblong 
to sagittate or linear, often more or less connivent around 
the filiform style with its small, generally simple stigma, 
or at least erect, very rarely distant; flowers generally 
conspicuous and highly colored. 

This is the type upon which my interest has been for 
the most part centered. It shows a great uniformity of 
character and in this lie its especial interest and impor- 
tance as a source of data towards the solution of the prob- 
lem of the influence of insects upon the form of flowers 
and the geographical distribution of different floral types. 
In view of this fact, the treatment of this class is more 
detailed than that of the others. 

As I have insisted above, the distinction between dehis- 
cence by pores and by longitudinal slits, and between the 



182 MISSOURI BOTANICAL GARDEN. 

Solanum-C&ssia type and any other is not an absolute one. 
The number of apically dehiscent genera or species might 
be easily increased or decreased by including forms in 
which the lateral slits first open more widely at the tip, or 
including all those in which the pores are finally supple- 
mented by lateral slits at length continued more or less down 
the side. The number of genera as I have limited it is, I 
feel confident, approximately right so far as may be deter- 
mined from systematic literature and the examination of 
herbarium material. 

To the Solanum-Cassia type I have assigned 59 genera. 
Of these, 19 are Monocotyledons and 40 Dicotyledons. The 
Monocotyledons are: JMayaca (Mayacaceae), ScJioenoce- 
phalium, Stegolepis, Rapatea, Saxo-Fridcricia, Gephalos- 
temon, SpatJianthus (Rapateaceae), Oarlonema, Dichori- 
sandra (Commelinaceae), 3£onochoria (Pontederiaceae), 
Walleria, A.grostocrinum, Dianella, Galectasia, Luzuri- 
aga (Liliaceae), Oonanthera, Qyanella, Zephyrd, Teco- 
pliilea (Amaryllidaceae). The Dicotyledons are : Gheiraa- 
ihera (Pittosporaceae), Cassia, ICoompassia, Distemonan- 
thus, Labichea, Dicorynia, Baudouinia, Duparquetia, 
ITrameria, Marliusia (Leguminosae), Platytheca, Tetra- 
theca, Tremandra (Tremandraceae), Thomasia, Guiche- 
notia, Lysiosepalum, Lasiopetalum (Sterculiaceae), Oura- 
tea, Brachenridgea , G-odoya, Elvasia, Blastemanthus , 
Wallacea, Schuurmansia , Poecilandra, Luxembergia, 
Euthemis, Leitgebia (Ochnaceae), Stemonoporus , JSfono- 
porandra (Dipterocarpaceae), Kiggelaria (Flacourtiaceae), 
Begonia/, sections Solanthera and Parvibegonia (Begonia- 
ceae), Ardisia, sections Icacorea, Stylogyne, and Monopo- 
rus (Myrsinaceae), G-ardneria (Loganiaceae), Exacum, 
Gotylanthera ( Gentianaceae ) , Solarium, Cyphomandra 
(Solanaceae), Argostemma, and Strumpfia (Rubiaceae). 

Mayacaceae. 
In the Mayacaceae, with the single genus Afayaca, we 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 183 

have a very good representative of this type, although the 
anthers offer some peculiarities of structure. (Ten species; 
all but one, which is found in Lower Guinea, are American ; 
1 very similar to and possibly identical with a South Ameri- 
can form occurs in eastern North America, 1 occurs in 
Cuba, and the others in Brazil, Peru and Guiana. Brazil 
has six species.) 

Rapateaceae. 

The Rapateaceae is a family of six genera and about 19 
species in tropical South America. Its position in this 
group is somewhat questionable on account of the moder- 
ately long perianth tube formed by the lower portion of 
the segments. The limb, however, is patent, the anthers 
basifixed and linear and shedding their pollen through 
terminal, pores or a single terminal pore, sometimes pro- 
vided with a terminal, spoon-like prolongation of the tip of 
the anther. 

Commelinaceae . 

The Commelinaceae is represented by two genera besides 
the highly interesting Oochliostema to be described later. 
Oartonema (5 or 6 species confined to tropical Australia) 
shows a considerable range of form in the structure of the 
anther. In some species, dehiscence is by longitudinal 
slits; dehiscence by pores seems to be in a less perfected 
state than in the tropical American Dichorisandra (about 
27 species, almost exclusively Brazilian). This genus is a 
highly interesting and very characteristic representative of 
the Solanum-Cassia type. 

Except for minor differences, a considerable degree of 
uniformity prevails in the floral structure of the Comme- 
linaceae. The petals are patent, except in about 3 genera 
in which they are unguiculate and united into a usually 
narrow tube, and 1 in which the perianth is tubular 
below, but even here the limb is generally patent. The 
basifixed anthers show a wide range of form, while the 



184 MISSOURI BOTANICAL, GARDEN. 

filaments may be naked or provided with hairs. Thus, 
excepting the character of dehiscence, with perhaps a slight 
elongation of the anthers, the flowers of the two forms 
treated here do not differ widely from the other genera of 
the family. In floral structure, the most highly organized 
member is <3 ocliliostema , to be discussed later. 

Pontederiaceae . 

Monochoria (3 species, 1 in tropical East Asia, 1 in 
tropical and subtropical Asia and East Africa, and 1 in 
Australia), of the Pontederiaceae, has the corolla and the 
form of the anthers of this type. The filaments are not so 
reduced in length as is generally the case, and the terminal 
pores are soon continued down the sides in longitudinal 
slits. 

A comparison of the disposition and form of the stamens 
in their relation to the length and form of the perianth 
tube in the other genera of this family is interesting. In 
Ueteranthera the perianth tube may be very long and nar- 
row, but the limb is widely patent. In the other genera, 
the perianth is more widely expanded from the base but 
still shows a tube of considerable length. Only in ^Monoch- 
oria is the tube wanting and the elements of the perianth, 
which are almost free to the base, widely patent. Con- 
cerning the form of the stamens it may be said that in 
Eichomia, Pontederia and Heussia, the markedly zygo- 
morphie androecium is characterized by filaments of con- 
siderable length with basifixed or nearly basifixed and short, 
thick anthers which may be included or widely exserted, 
In Ueteranthera the anthers are sometimes more elongate, 
approaching linear in form. The filaments, inserted at the 
top of the corolla tube, may be of the same length as the 
anthers or much longer. In Monochoria, the filiform fila- 
ments are of about the same length as the anthers, which 
seem to be the most elongate in the family, and, perhaps, 
with walls of firmer texture. 



THE DEHISCENCE OP ANTHERS BY APICAL POKES. 185 

Liliaceae. 

Of the Liliaceae, five genera may be mentioned : 
Walleria (three species, or perhaps only three varieties of 
the same species, in tropical Africa, with 1 extending into 
South Africa, and another from Madagascar) is one of the 
finest illustrations of this type which I have seen. A.gros- 
tocrinum (1 Australian species) is another good represen- 
tative with somewhat zygomorphie flowers. Dianella (11 
species, 1 widely distributed in the Mascarine Islands, 
tropical Asia, Australia, New Caledonia and the Sandwich 
Islands, 8 more in Australia, 1 in New Zealand, Norfolk 
and Fiji and Society Islands and 1 in the Isle of Pines and 
New Caledonia), with variously thickened filaments and 
anthers in which the pores are sometimes continued down- 
ward in introrse longitudinal slits, also belongs here. The 
flowers of Calectasis (1 West Australian species) are 
among the most beautiful of those assigned to this type. 
Luzuriaga is a highly interesting genus of South Pacific 
or antarctic distribution. Four species are described. In 
three species (of which 1 is confined to New Zealand, 1 
occurs in Patagonia, Tierra del Fuego, Falkland Islands and 
New Zealand, and 1 is found only in Chile and Peru) the 
flowers have a very close similarity to other members of 
the Solanum-Cassia type, but dehiscence seems to be by 
longitudinal slits, although it has been described as by 
pores and it may be so at first, as in Monochoria and vari- 
ous other forms. The fourth species (reported only from 
South Chile) has a different floral structure. The anthers 
are borne on longer filaments and at the time of flowering 
are reflex ed and open, at least at first, by two basal pores. 
The reflexed stamens are quite closely approximated 
around the ovary so that the general habit of the flower 
agrees very closely with that of many other forms in this 
class. 

Odonlostomum is an aberrant form from California which 
will be considered later. 



186 MISSOURI BOTANICAL GARDEN. 

Amaryllidaceae. 

Pax in his treatment of the Amaryllidaceae in Die 
Natiirliohen Pflanzenfamilien, characterizes two tribes, 
Galanthinae and Conanthereae, as dehiscing by apical pores. 
The genera of the Galanthinae havebeen mentioned under 
the Ericaceous type. 

Conanthera (3 or 4 species in Chile) has the anthers 
connivent in a cone and dehiscing by pores at the tip or 
by introrse slits extending almost to the base . They are 
provided at the tip with a simple or bifid acumen and are 
either exserted or included in the eampanulate perianth 
tube. Gyanella (4 or 5 species in the Cape region) is a 
beautiful example of this type with some species showing 
zygomorphic flowers, in some respects very similar to those 
of Cassia. Zephyra (1 species in Chile) and Tecophilaea 
(2 species in Chile) are aberrant, zygomorphic forms 
which cannot be described in detail here. 

A point of interest in the relation of the apically dehis- 
cent genera of the Liliaceae and Amaryllidaceae to the. 
other members of the families is that in the arrangement of 
Engler and Prantl the apically dehiscent genera of the for- 
mer are found associated with sometimes several of the 
longitudinally dehiscent forms in as many of the ultimate 
groups as there are genera, belonging to three of the sub- 
families of that system, while in the Amaryllidaceae, the 
four genera assigned to the Solanum-Cassia type all belong 
to the same ultimate group. 

A comparison of these groups in the two families is 
interesting. In the Amaryllidaceae the four apically 
dehiscent genera belong to a single group and are struc- 
turally very similar, differing largely in the characteristics 
of the variously zygomorphic androecium. In the Lili- 
aceae, however, the differences to be noted within the 
same group are much greater, as may be most strikingly 
illustrated by the comparison of Walleria with Oloriosa 



THE DEHISCENCE OP ANTHERS BY APICAL PORES. 187 

Sandersonia, and Tricyrtis, while other members of the 
group also show considerable differences in structure, and 
none of them at all closely approach the form of the 
apically dehiscent genus. In the twenty genera of the 
Anthericinae, the group to which A.grostocrinum belongs, 
however, the floral habit is quite uniform. In all, the 
corolla is rotate ; in several, the filaments are long and the 
anthers short, but in others the filaments are much reduced 
in length and the anthers oblong to linear and sometimes 
tending to be more connivent around the style. 

Space will not permit a detailed discussion of the floral 
structure of the two families. The two illustrations just 
given from the Liliaceae are perhaps representative. I 
think it may be said for these families that in the Ama- 
ryllidaceae the prevalent habit of the perianth is, roughly 
speaking, infundibuliform while in the Liliaceae it is 
patent or campanulate. In the Amaryllidaceae the four 
apically dehiscent genei'a stand almost alone as representa- 
tives of this floral habit while in the Liliaceae there are 
many genera with patent or broadly campanulate perianth. 
In'the most of these genera the filaments are filiform, more 
or less elongate, and bear short, usually versatile anthers, 
but in several the filament is reduced in length while the 
anther is of the more elongate form commonly seen in the 
apically dehiscent genera. The forms which in their general 
structure bear a close resemblance to the apically dehiscent 
Solanum-Cassia type are almost wanting in the Amaryllid- 
aceae. It must not be understood that they are identical 
with the apically dehiscent genera except for the difference 
in the method of the opening of the anthers. I do not go 
so far as to state that tbey represent incipient stages of 
members of this class, for in the interpretation of such 
facts as these, the greatest caution must be exercised. It 
must be pointed out, however, that in the general habit of 
the flower there is not the sharp line of distinction be- 



188 MISSOURI BOTANICAL GARDEN. 

tween apically and longitudinally dehiscent forms in the 
Liliaceae that there is in some other families. 
We may- now pass to the Dicotyledons. 

Pittosporaceae . 

We find in Solly a (2 Australian species) of the Pitto- 
sporaceae a form similar in structure to the section Lyco- 
persicum of Solarium, but not showing dehiscence by pores, 
and Cheiranthera (4 Australian species) which is an excel- 
lent, slightly zygomorphic representative of this type. 

In the Pittosporaceae the five petals, frequently more or 
less unguiculate, are erect and connivent or coherent in a 
tube at the base and spreading above. In only a few 
species besides those of the genera Sollya and Cheiran- 
thera are the petals more or less patent from near the base. 
Another most important difference between these two 
genera and the others of the family is the relative size and 
proportion of the parts of the stamen. The anthers are 
longer than the filaments in Sollya and Cheiranthera, 
while in the other genera they are shorter than the fila- 
ments, and, in relation to other parts of the flower, smaller 
than in the genera considered here. These two genera are 
aberrant forms in the family. 

Leguminosae . 

In the Leguminosae the apically dehiscent genera are, 
with one exception, confined to the Caesalpinioideae, and 
in this sub-family, with one exception, to the Cassieae. 

Cassia, serving as one of the types of this group, is too 
well known to require description. (About 412 species of 
tropical and subtropical distribution are known, extending 
in America from Patagonia into the United States and 
reaching their highest differentiation in the tropics of the 
New World — 290 species. Australia has about 33 species, 
tropical Africa 27, the Indian, Malayan and Oceanic Island 
region about 34, and the Cape region 5 or 6). 

In lEoompassia (represented by 2 or 3 species, gigantic 



THE DEHISCENCE OF ANTHER8 BY APICAIi PORES. 189 

trees, confined to Malacca and the Malay Archipelago) the 
stamens of the species differ considerably in form, those 
of one species being much broader than those of the other, 
and possibly in dehiscence. 

DistemonantJius (1 species found in Upper Guinea) is a 
strongly zygomorphic form with only two fertile stamens, 
clearly belonging here. 

Labichea (5 Austrialian species) is very closely related 
to Cassia and with almost identical floral structure except 
that the stamens are reduced to two. 

Storclciella has the perianth quite widely patent and the 
linear anthers of this type, but the elongate filaments make 
it necessary to place it in the Melastomataceous type. 

Dicorynia (4 species in Guiana and North Brazil), with 
its two short, thick anthers, one of which is sometimes 
eight-locellate at the tip, is a unique form, but one which 
seems best treated here. 

Saudouinia (2 species in Madagascar) has linear-sagit- 
tate, basifixed, apically acuminate and penicillate anthers, 
both locules of which open at first by an introrse subapical 
fissure which soon extends in two introrse longitudinal slits 
to the base. 

In Duparquetia (a single species, a richly flowering 
shrub, in west tropical Africa) the flowers are, with the 
possible exception of ICrameria, the most strongly zygo- 
morphic in this type. The four anthers dehisce by short, 
terminal slits which are not continued down the side for 
more than a third of the whole length of the anther. 

Marlinsia (2 species in Brazil and British Guiana) is 
one of the largest-flowered and most typical, slightly zygo- 
morphic, representatives of this type. 

Krameria (23 species, distributed from warmer North 
America to Chile), of somewhat uncertain systematic 
affinities, is now placed next to the Cassieae in the Legumi- 
nosae. Structurally it is the most aberrant form assigned 



190 MISSOURI BOTANICAL GARDEN. 

to this type, but it may be treated here, provisionally, at 
least. 

In the Leguminosae, the apically dehiscent forms of this 
type are confined, with the single exception of Krameria, 
to the Cassieae of the Caesalpinioideae. The genera of 
the Cassieae are distinguished from all others of this sub- 
family except Krameria by their usually basifixed, apically 
dehiscent anthers, those of all other genera being dorsifixed 
and versatile and with longitudinal dehiscence. In exam- 
ining the genera of the Cassieae itself I find that the apically 
dehiscent forms have, as compared with the others, a more 
elongate anther and a shorter filament. The patent corolla 
is general in the Caesalpinioideae but the form of the anthers 
in the apically dehiscent genera will, I think, be found 
quite different from that of the other genera. 

Tremandraceae . 

The Tremandraceae (represented by 3 genera of about 
23 species endemic in West and South Australia) constitute 
the only dicotyledonous family in which all the species are 
apically dehiscent. The four parallel cells in the same 
plane in the anthers of JPlatytheca, and the terminal tube 
opening by a single pore in this genus and Tetratheca merit 
especial mention. The genera of this family, while offer- 
ing some minor points of difference from the others, must 
be regarded as among the most highly specialized of this 
type. 

Sterculiaceae. 

In the Sterculiaceae, JETermannia shows dehiscence by 
longitudinal slits, but the habit of the androecium is so 
similar to that of Solanum, Borago and some other forms 
that it may be mentioned for comparison. 

In Thomasia (21 species, all but 1 confined to Australia) 
mu'ch the same condition prevails except that here the 
ovate or oblong anthers, connivent in a cone around the 
ovary, dehisce by short, sometimes almost poriforai,introrse 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 191 

slits, finally extending more or less down the sides. The 
pistil sometimes considerably exceeds the tips of the 
stamens. 

Quichenotia (5 species endemic in West Australia) has 
anthers opening by introrse terminal pores or slits which 
may be continued down the side. 

Lysiosepalum (2 species are found in West Australia) 
has the anthers subulate to linear and dehiscing by apical 
poriform slits which may be finally continued down the 
sides. 

Lasiopetalum (25 Australian species, mostly confined to 
West Australia) is probably the best representative of 
the Solanum-Cassia type to be found in the. Sterculiaceae. 
Even here the pores of the anthers are continued down the 
sides and the style is sometimes much exserted and covered 
with stellate hairs. 

It is unnecessary to discuss in detail here the polymor- 
phic and often complicated floral organization in the Ster- 
culiaceae. The forms showing dehiscence by pores or 
short slits are found only in the Lasiopetaleae, which dif- 
fers very essentially from the other tribes in its floral 
structure. All members of this group agree in the posses- 
sion of a patent perianth. The petaloid nature of the 
sepals and the reduction of the petals to scale-like struc- 
tures, the peculiar style in some species and the floral habit 
of other genera of the family to which these forms have 
been assigned give them a peculiar interest, but our knowl- 
edge of them is entirely too meager to permit of any 
suggestions as to the significance of the forms. 

Ochnaceae. 

In the Ochnaceae, we have a family particularly difficult 

of treatment on account of the numerical reduction in the 

androecium. Transitions are present between the Dilleni- 

aceous and the Solanum-Cassia type, while some of the 



192 MISSOURI BOTANICAL GARDEN. 

forms are strongly suggestive of the Melastomataceous 
type. 

Ouratea (76 species in tropical America and 26 in the 
Old World) is one of the finest examples of the Solanum- 
Cassia type, sometimes with more or less rugose anthers. 
JBrackenridgea (5 species, 2 in the Fiji Islands, 1 inPenang, 
1 in Queensland, 1 in Zanzibar) is an excellent represen- 
tative of the floral habit of this type, but the anthers open 
by longitudinal slits or at first by apical pores which are 
later increased more or less towards the base in longitudi- 
nal slits. Crodoya (3 species in Peru and New Granada) has 
10 to 20 stamens but their form is such that they seem to 
be more properly treated here than in the Dilleniaceous type. 
Elvasia (4 species in Brazil and Guiana) with stamens 8 or 
indefinite, up to 20, with oblong or nearly oval anthers 
basifixed on more or less elongate, filiform filaments, sug- 
gests in some of its species the Melastomataceous type, but 
in others the filaments are reduced in length and the anthers 
more elongate, so it may be treated here. 

Blastemanthus (2 species from the upper Amazon, of 
which 1 also occurs in Guiana) has the 10 declinate sta- 
mens with very long anthers on short filaments surrounded 
by numerous staminodia. As far as form is concerned, 
Wallacea (1 species in the upper Amazon region) is an 
excellent representative of this type. The fertile stamens 
are 5 in number and turned to one side at the time of flow- 
ering. Staminodia are present, and the terminal pores, or 
single terminal pore, of the anthers are said to be later 
continued down the sides in lateral slits. In Schuurmansia 
(3—4 species in the Indian Archipelago) the floral struc- 
ture seems to be very similar to the preceding. Poecilan- 
dra (1 species in northern Brazil and British Guiana) may 
be placed without question in this type. Luxembergia 
(7 species, all in Brazil) is a peculiar genus in which the 
indefinite, usually few, anthers are aggregate or connate in 
a column on one side of the ovary. Eulhemis (3—4 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 193 

«pecies in the Indo-Malay Archipelago) is a good example 
of this type. 

The numerical reduction in the androecium of the Och- 
naceae may be mentioned in this connection. The family 
is characterized by elongate, basifixed anthers, generally 
dehiscing by apical pores, with some species of several of 
the genera showing, at least finally, longitudinal slits. 
I have been unable to convince myself of any unquestion- 
able relation existing between the stage of reduction in the 
number of stamens and the perfection of dehiscence by 
pores in this family except in the following special cases. 
"W hile too much weight must not be attached to the limited 
evidence, the condition of stamens and staminodia in cer- 
tain genera may be described. 

In JVecJcia (3 species in the Indo-Malay Archipelago) 
-the staminodia of the inner whorl, about 10 in number, are 
clavate and cohere with the filaments of the longitudinally 
dehiscent anthers at the base. In Leitgehia (1 species in 
the savannahs of Brazil) a similar condition occurs with only 
1 whorl of 5 spatulate staminodia and anthers dehiscing 
at first by apical pores. In Sauvagesia (11 species, 
confined, with the exception of 1 of universal tropical 
distribution, to tropical Brazil) two whorls are present, the 
outer of filiform staminodia, the inner of 5 petaloid struc- 
tures, closely approximated in a cylinder around the longi- 
tudinally dehiscent anthers. In Lavradia (6 species ende- 
mic in Brazil) is to be seen a still more modified condition. 
The outer whorl of staminodia is entirely wanting while the 
members of the petaloid whorl are connate in a coniform or 
almost urceolate corona surrounding the shortened anthers 
which seem to show no trace of apical pores, but open 
longitudinally from the first. These aberrant forms may 
be profitably compared with Antholoma of the Elaeocar- 
paceae. While the evidence is far from comprehensive, it 

13 



194 MISSOURI BOTANICAL GARDEN. 

strongly suggests the significance of the exserted condition 
of the anthers in apical dehiscence. 

Dipterocarpaceae . 
In the Dipterocarpaceae, species of Shorea, A.nisopiera 
and Vatica have anthers opening more or less terminally, 
but the only genera to be considered in this paper are 
Stemonoporus (12—13 species endemic in Ceylon), and 
Monoporandra (2 species endemic in Ceylon), both of 
which in the general habit of the flower fall clearly into 
the Solanum-Cassia type. 

Flacourtiaceae. 
Of the Flacourtiaceae, ITiggelaria (3 South African 
species) may be assigned to this type. 

Begoniaceae. 
In the Begoniaceae, ^Begonia is the principal genus, being 
represented by about 400 species throughout the warmer 
regions of the whole world. In all the species , the stamens 
are numerous, anthers basifixed, rarely almost spherical, 
usually ovate or oblong to linear, connective produced in 
various forms beyond the locules or not, filaments long or 
short as compared with the anthers, free or monadelphous. 
The dehiscence is by lateral slits. In two Brazilian species, 
forming the section Solanthera, the filaments are free, 
anthers linear, obtuse, much longer than the filaments and 
opening at the apex through two pores. In the section 
Parvibegonia, of 8 Indian species, dehiscence is said to be 
lateral, by short, subapical slits in six species, and by 
"pores" in the other two. My knowledge of the Indian 
species is very incomplete, but the Brazilian forms may 
certainly be placed in this type. 

Myrsinaceae. 
In the Myrsinaceae, Qybianihus, with small broad an- 
thers dehiscing by elongate or short and subapical, some- 



THE DEHISCENCE OP ANTHERS BY APICAL PORES. 195 

times almost porif orm slits, is an interesting form but hardly 
merits consideration as one in which dehiscence is by apical 
pores. 

Ardisia is a large genus of over 200 species much in need 
of revision. The stamens are 5 in number, with short 
sagittate or lanceolate anthers usually opening by longitu- 
dinal slits. In the section Icacorea, however, according to 
Pax, the anthers open by apical pores. This section seems 
to comprise about 7 South American species. The anthers 
of the second section, Sty logy ne, with 1 Brazilian and 1 
Malayan species, are described as opening by terminal 
pores. In the single representative of the third section, 
A., paludosa of Madagascar, the anthers open by a single 
terminal pore. In the two other sections of the genus 
containing the remainder of the species, distributed 
throughout the tropics of both hemispheres, dehiscence is 
by lateral slits. My knowledge of this genus is very im- 
perfect, but I think it probable that an examination of 
suitable material would show that in the most of the above 
species described as apically dehiscent the pores are not 
permanent, but merely the beginnings of longitudinal 

slits. 

Loganiaceae. 

In the 32 genera assigned to the Loganiaceae only 5 or 6 
have any species with a rotate corolla, the tendency being 
towards tubular, salver-form or campanulate. Gardneria 
is the only genus characterized by a rotate corolla with 
oblong to linear exserted anthers. Three species are 
found in Japan and India. In Gf. nutans (of Japan) the 
slits are said not to extend to the base as in the others, but 
to be confined to the tip of the anther. 

Gentianaceae. 
In the Gentianaceae, two genera belong clearly to this 
type: JExacum (29 species, distributed over tropical and 
sub-tropical Asia, the Malay Archipelago, Madagascar and 



196 MISSOURI BOTANICAL GARDEN. 

tropical Africa, with 4 species in tropical America) , and 
Gotylanthera (3 species, saprophytic herbs, from Java, 
Mariana Islands and the Himalayas). Qhironia and JDeia- 
nira are interesting for comparison, though they can hardly 
be included in the list. 

In the Gentianaceae the corolla is infundibuliform, hy- 
pocrateriform, campanulate or sometimes rotate. In both 
Exacum and C otylanthera the corolla has a short, subglo- 
bose tube with patent limb and stamens inserted in the 
throat. A detailed comparison of the floral habit of these 
genera with the other members of the family is precluded 
by lack of space. Several genera approach the rotate co- 
rolla characteristic of the Solanum- Cassia type either by 
way of a broadly campanulate corolla or by a hypocrateri- 
form corolla with a usually shortened cylindrical tube and 
prominent patent limb. In some of these forms the sta- 
mens are included while in others they are exserted, but 
with short, versatile anthers on longer filaments. In some 
cases the anthers are linear but versatile, while in others 
they are linear and basifixed. A few forms approach very 
closely in their floral habit the two apically dehiscent 
genera. 

Solanaceae. 

In the Solanaceae, two forms are to be considered, Sola- 
num and Oyphomandra. Solatium (of about 970 nomi- 
nal species 630 occur in Tropical America, 70 in extra- 
tropical South America, 27 in tropical Africa, 10 in the 
East African islands, 34 in the Indian, Malay and Oceanic 
Island region, 26 in South Africa and 52 in Australia) is 
too well known to require description: attention may 
be directed to the few zygomorphic forms constituting 
the section Nycterium, and to the considerable number of 
species showing incipient stages of this characteristic, and 
to the condition prevailing in the small section Lycopersi- 
cum. In many of the species the pores are later continued 
down the sides of the anthers, sometimes to the base, in 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 197 

longitudinal slits. Solanum furnishes one of the classic 
illustrations of the apically dehiscent anther, and the large 
number of species in which the pores are continued down 
the sides in longitudinal slits stands as a justification for 
including in the apically dehiscent category several genera 
in which the pores are later supplemented by lateral slits. 
Between the two forms of dehiscence, no sharp line exists 
although in the individual cases it is usually not difficult 
to decide very satisfactorily to which category a form 
should be referred. 

In Cyphomandra (about 35 tropical American species, 
especially numerous in Brazil, and 2 extratropical South 
American species) , the species of which were formerly 
treated under Solarium, the structure of the anther differs 
somewhat from that of Solanum. In some of the species 
at least, the walls are very thin and elastic, so that the 
pollen is puffed out by a bellows-like action. This genus 
is, in my present judgment, to be regarded as a specialized 
representative of the Solanum-Cassia type. 

In the Solanaceae the gamopetalous corolla is tubular, 
infundibuliform, hypocrateriform, campanulate or rotate. 
I shall not give here a detailed discussion of the structure 
or distribution of the genera or species which approach 
Solanum in floral habit, but from quite careful comparison 
I may state with some confidence that Solanum represents 
the form in which the anthers have attained the greatest 
length and the filaments are the most reduced. 

Kubiaceae. 

In Argostemma (30—40 species in tropical East Asia, 1 in 
"West Africa) , of the Rubiaceae, we have an interesting genus, 
some species of which belong clearly to this type. Only a 
few of the 30—40 species show dehiscence by pores, and I 
have not data which enable me to state which or how many 
these species are. 

N'eurocalyx (about 6 species especially in Ceylon, but 



198 MISSOURI BOTANICAL GARDEN. 

also extending to the island of Borneo and oriental Asia) 
is an interesting form for comparison, in which the anthers 
are connate in a cylindrical tube and open internally. The 
monotypic Strumpfia, of the rocky coast of the Antilles, is 
another form which may be profitably compared. The five 
anthers form a conical synandrium around the pistil. The 
locules open in a more or less irregular manner at the tip, 
the outer wall of the synandrium extending considerably 
above the inner one. To attempt a statement of the con- 
dition of the corolla, androecium, and gynoecium in the 
genera of this family, even were data for such a statement 
available, demands an unprofitable amount of space and 
time. A quite careful examination of the genera, however, 
has shown that in the most of those with a rotate or almost 
rotate corolla, the androecium is characterized by elongate 
filaments and short anthers. In some cases the anthers 
may be elongate or linear, but in these cases they are 
usually dorsifixed, sometimes near the base, and distant 
instead of connivent around the style. With the exception 
of A.vgostemma, JSTeurocalyoc and Strumpjia, I have found 
no genera which I could confidently refer to the floral 
habit exemplified by Solanum or Cassia. All of these 
forms may be profitably compared with the section Lyco- 
persicum of Solarium or with Solly a of the Pittosporaceae. 

VII. MELASTOMATACEOUS TYPE. 

This type is practically coextensive with the Melastoma- 
taceae, which furnishes almost all of its members. Possi- 
bly some of the genera from this family should be 
included under the Solanum-Cassia type on purely struc- 
tural grounds, but considering the nature of the character- 
istics separating the two groups, it has been thought best 
to retain all such forms here. 

The essential distinguishing characteristic of this type is, 
in my opinion, the elongated filaments. There is the same 
conspicuous, patent perianth as in the Solanum-Cassia type 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 199 

with little indication of a special receptacle for nectar. 
The number of stamens is usually small. The anthers are 
basifixed and have generally the same elongate, linear or 
subulate, form as in the preceding type. The pores are 
often minute, both locules usually opening through a single 
pore, and are very rarely continued down the sides. Often 
the anther has thin, flexible walls which make possible a 
bellows-like action in the ejection of the pollen. 

In the Melastomataceae, considering the size of the fam- 
ily, the floral structure is quite uniform. Flowers usually 
4- or 5-merous ; petals- usually large and highly colored, 
never absent though sometimes reduced, rarely connivent 
into a tubular or campanulate corolla; stamens usually 
twice as many as petals, rarely as many, or in some cases 
indefinite, all similar or alternately reduced or imperfect; 
filaments long or short, thick or slender, straight or arcu- 
ate, glabrous or glandular ; anthers basifixed , almost spheri- 
cal or cuneiform to linear or subulate, the oblong to linear 
or subulate being the prevalent type, straight or variously 
curved, dehiscing by a single terminal pore, rarely by two 
terminal pores and very rarely by lateral slit-s or by four 
terminal pores. The connective shows many modifications 
and is often produced below the locules and variously ap- 
pendaged. Style filiform, straight or curved; stigma 
usually minute, rarely capitate. 

It seems unnecessary to enumerate or describe the 
genera. Tlie more essential structural points and the 
geographical distribution of the forms may be obtained 
from the work of Bentham and Hooker or Engler and 
Prantl, the splendid elaboration of the South American 
forms in Flora Brasiliensis or the detailed monograph of 
the family by Cogniaux. An examination of the indi- 
vidual genera reveals the fact that there prevails in this 
family a marked structural uniformity which instead of 
being obscured by numerous minor differences, is only 
made more conspicuous by them. The family includes 



200 MISSOURI BOTANICAL GARDEN. 

161 genera of tropical and subtropical distribution, 98 
occurring in South America, 31 in the Indian region, 1& 
in the tropical African region and the others of various 
distribution. 

Outside of this family I have assigned only three genera 
to this type : Storckiella of the Leguminosae, and Max- 
imilianea and Aimoreuxia of the Bixaceae. Possibly some 
others, as, for instance, OJieiranthera of the Pittosporaceae 
and Exacum of the Gentianaceae should also have been 
placed here instead of in the Solanum-Cassia type. 

Storckiella (2 species in Oceanica) is the only member 
of the apically dehiscent Cassieae not clearly belonging in 
the Solanum-Cassia type. It is the only form with long 
filaments and it seems best for this reason to separate it 
from the genera to which it is systematically related. 

In the Bixaceae, Bixa has been treated under the Dillen- 
iaceous type. 3£aximilianea (13 species, 6 in tropical 
America, 3 in Africa, 1 in tropical Africa and southern 
Asia, 3 in North Australia and Queensland) with large 
actinomorphic flowers, and A.moreuxia (3 species in Central 
America) with zygomorphic flowers, have long filaments 
and rather elongate anthers which open by well-formed 
terminal pores. These forms should possibly have been 
placed in the Dilleniaceous type too, but the form of the 
stamens is very suggestive of the Melastomataceous type 
so they have been placed here. 

ABERRANT FORMS. 

As mentioned above, a few genera in which the anthers 
have been described as opening by pores do not fall readily 
into any of the above types. The fact that the most of 
the forms described as apically dehiscent belong clearly to 
one of these categories has led me to designate these simply 
as aberrant forms. Here as elsewhere classification should 
not be too much influenced by the number of individuals or 
named groups of individuals which may be assigned to any 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 201 

particular subdivision, but in our alroost total ignorance of 
their ecology and in some cases of their structure it seems 
best to treat these forms in this way. 

These so-called aberrant forms are the following : — 

Odontostomum (1 species in California) of the Liliaceae 
has a hypocraterif orm corolla with patent or reflexed lobes 
with erect, basifixed, ovoid anthers which are only about 
one-half or one-third as long as the filaments. 

In the Euphorbiaceae, Poranihera (5 Australian species, 
of which one also occurs in Tasmania and New Zealand, 
and a sixth endemic in New Zealand) has small flowers with 
five stamens having 4-locellate anthers opening in four 
terminal pores which are quite distinct or sometimes con- 
fluent into two . 

In the Ebenaceae, the anthers of the three species of the 
section Leucoxylum (2 in Madagascar and 1 in the Indian 
monsoon region) of Diospyros (about 120 species of wide 
distribution, especially numerous in the Indian region) are 
described as opening by lateral pores at the tip instead of 
by longitudinal slits as in the other species. The corolla 
in Diospyros is urceolate, campanulate, tubular, or salver- 
form ; stamens 4 to indefinite, usually 16; anthers oblong, 
linear or lanceolate. In the three species which have been 
described as apically dehiscent the flowers are small, the 
corolla seems to be openly campanulate and the stamens 
10—16 in number. It may be that these species might be 
properly assigned to the Solanuin-Cassia type. Sargent 
describes the sixteen anthers of D. Texana included in the 
urceolate corolla as dehiscing only near the apex and his 
figure represents the openings as almost poriforrn. 

The condition prevailing in the Acanthaceae is one of par- 
ticular interest. Here we have 7 genera which are said 
to have, in some species at least, anthers opening by apical 
pores. Of these, Staurogyne need not be considered, since 
dehiscence in this genus can hardly be justly described as 
by apical pores. Ophiorrhiziphyllon (1 species in Marta- 



202 MISSOURI BOTANICAL GARDEN. 

ban) is insufficiently known to me. The anthers of Afro- 
mendoncia (3 species in tropical Africa) are peculiar 
among those opening by apical pores. In Hiernia (1 
species in Angola), JHendoncia ("about 20 species in trop- 
ical America), Monachochlamys (1 species in Madagascar) 
and Pseudocalyx (1 species in Madagascar and Nossi Be 
Island) the anthers are of the simple oblong to linear type 
with the more or less shortened filament so characteristic 
of the apically dehiscent anthers of most of the preceding 
families. Except in one or two of these six forms, how- 
ever, dehiscence is not by the clearly-defined, round pores 
found in so many of the genera described above, but by 
more or less elongate apical slits. 

It is not necessary in this place to enter into a detailed 
discussion of the structure of the stamens in the usually 
strongly zygomorphic flowers of the Acanthaceae, but at- 
tention may be called to the fact that the form of stamens 
found in the four genera just mentioned is not at all com- 
mon, in fact is almost unique, in this large family. In 
almost all the forms discussed above, the linear anthers on 
short filaments are exserted from a widely open corolla . 
In the Acanthaceae the corolla is rarely open enough to 
be designated as campanulate from the base and it is never 
rotate. In Hiernia the limb is patent and the somewhat 
declinate linear anthers exserted so that except for the con- 
siderable length of the tube the habit of the flower is much 
the same as if the anthers were inserted by short filaments 
on a rotate corolla. The presence of the long perianth 
tube alone prevents the placing of this interesting genus in 
the Solanum-Cassia type. In Ophiorr7iiziphyUon the two 
fertile stamens with long filaments and short anthers are 
described as much exserted from the zygomorphic corolla. 
A-fromendoncia has the tube somewhat expanded above 
and the characteristic anthers project only slightly. In 
3fonachoc7damys the corolla is tubular to campanulate, 
with the stamens clearly included, while in the two other 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 203 

genera, Mendoncia and Psetcdocalyx, the same condition 
prevails with a less widely open corolla tube and more 
elongate anthers. Data are too limited to justify specula- 
tion, but the association of a tubular corolla with apically 
dehiscent anthers of the form so generally found when the 
corolla is patent, is interesting. The inclusion of a linear 
anther in a cylindrical corolla tube is nothing out of the 
ordinary, but the cases detailed above are the only ones in 
which the dehiscence is by pores. The condition described 
in certain aberrant Elaeocarpaceae and Ochnaceae will be 
recalled in this connection. 

In contrast with the preceding family, the Rubiaceae has 
many genera and species in which the corolla is more or 
less widely open and several in which it may be character- 
ized as patent. Attention has already been called to some 
of these in the section devoted to the Solanum-Cassia type. 
Here we are concerned with only two genera. 

The anthers of Tresenthera (2 species, 1 in Venezuela, 
the other in the West Indies), somewhat exserted from the 
campanulate corolla, are rather unique in their form and 
mode of dehiscence, opening as they do by a triangular 
valve below the tip. In Mustia (5 species ranging from 
Central America to beyond the province of Rio de Janeiro 
in Brazil) the linear basifixed anthers are of the form so 
generally found in apically dehiscent genera. In some 
species the anthers are included in the tubular portion of 
the corolla while in others they are exserted their entire 
length from the throat while the rather large lobes of the 
limb are patent. The occurrence of linear anthers in this 
genus is not so worthy of comment as in Mendoncia, 
JPseudocalyx, and Hiernia of the Acanthaceae, where the 
anthers are generally very short, for in the Rubiaceae the 
anthers are frequently linear in form and included in the 
corolla tube. 

Too much significance must not be attached to the con- 
dition found in these genera, but it is suggestive when 



204 MISSOURI BOTANICAL GARDEN. 

compared with that of typical forms in the same family 
and with the typical and aberrant forms in apically dehis- 
cent groups. 

Systematic Relationship and Morphological Charac- 
teristics op Apically Dehiscent Forms. 

Dehiscence by pores may be characteristic of families, 
genera or species. In the Rapateaceae, Tremandraceae, 
and, with the exception of comparatively few genera and 
species, in the much larger Melastomataceae, dehiscence 
by pores is of regular occurrence. It may be found, how- 
ever, in only a few of the genera, as in Solarium of the Sol- 
anaceae, Cassia of the Caesalpinioideae, CheirantJiera of 
the Pitto,sporaceae and other examples which might be ad- 
vanced. Probably the best illustration is the monotypic 
Fissicalyx, the single apically dehiscent genus among over 
300 belonging to the Papilionatae. Only certain species 
of a genus, again, may have anthers with pores instead of 
longitudinal slits. The two Brazilian, and possibly two 
Indian species, of the 400 assigned to Begonia may be 
recalled in this connection, as may also three species 
forming the section Leucoxylum of JDiospyros. The api- 
cally dehiscent habit has unquestionably been assumed in- 
dependently by many unrelated genera and doubtless at 
various times. The regular occurrence of this method of 
dehiscence in several large groups indicates that in these 
cases the habit of dehiscence is much older, perhaps as old 
as the group itself. There is much to suggest that the 
apically dehiscent forms of the Solanum-Cassia and Melas- 
tomataceous types have originated from actinomorphic 
forms with more or less patent perianth. Evidences of 
exceptions to this generalization are, however, to be seen 
in several cases and in view of our almost complete igno- 
rance of primitive floral types we can hardly venture to 
hazard any statements as to the phylogeny of these forms. 

While evidence is too meager to permit of any conclu- 



THE DEHISCENCE OP ANTHERS BY APICAL PORES. 205 

sions as to the type of flower from which an apieally de- 
hiscent form of any of the classes here recognized may 
have been derived, it seems that there is little relation 
between the systematic affinities of a group and the 
possibility of its containing apieally dehiscent genera 
or species. 

STRUCTURAL, SIMILARITY OP MEMBERS OP THE SEVERAL 

TYPES . 

Structurally, the apieally dehiscent forms may be assem- 
bled into more or less natural and sharply limited cate- 
gories. The Araceous type is a natural and sharply 
limited one, distinguished by a uniformity of gross struc- 
ture that renders unnecessary any summary of morpholog- 
ical characteristics. The crowding of the anthers or their 
fusion into a synandrium renders apical dehiscence the 
most simple and the necessary mode, but the recognition 
of this fact does not furnish an explanation of the changes 
which have taken place in the development of the structure 
now prevailing or indicate the forces which have been 
active in producing them. Considerable is known of the 
pollination of these forms but upon the basis of the data 
which I have so far collected I am not yet prepared to 
discuss the significance of insects as a factor in the 
development of this floral type, or, to state the matter 
differently, to say in how far apical dehiscence in this 
type and the other structural characteristics correlated 
with (or determining) it are to be regarded as adaptations 
for fertilization by certain groups of animals. This 
much is certain, that the members of the Araceous type 
are distinct in every way from the others considered. 

This last statement also applies to the Gramineous type 
which may also be dismissed from our further considera- 
tion. 

The Polygalaceous type shows a great uniformity of 
structure. The locules of the short, broad anthers open 



206 MISSOURI BOTANICAL GARDEN. 

by a large terminal gap which may hardly be appropriately 
designated as a pore. These forms are evidently bee 
flowers in which the essential organs are inclosed in a 
ventral carina and this inclusion doubtless has a large 
significance in the form and size of the opening of the 
anther. 

The Ericaceous type is one of the most important, but 
demands for its adequate treatment much more space than 
can be given to it in this place and many more data than 
have so far been secured. The most conspicuous charac- 
teristic of this type is the pendulous habit of the flower, 
with tubular, urceolate or globose corolla. In this paper, 
several forms which differ in the essential features from 
the most prevalent characteristics of the type as I have 
limited it have been classed with this group largely on 
account of their systematic affinities. The class as a 
whole, however, may, I think, be regarded as a very 
natural one. 

The opening of the anther in this type varies greatly in 
form and size. The presence and form of the aristae on 
the anthers in many genera also furnish interesting char- 
acteristics and are of importance in the pollination of the 
flowers. 

The Ericaceae are known to be largely dependent for 
their fertilization upon Apidae, but the epitomizing of the 
observations on the ecology or the geographical distribu- 
tion of this large class lies outside the scope of the present 
paper. 

The Dilleniaceous, Solanum-Cassia and Melastomataceous 
types must be considered together. These types are not 
so sharply separated as the preceding but in some degree 
transgressive. It is this general group of floral forms to 
which the present paper is devoted. The others are con- 
sidered only in such detail as will show their relations, or 
rather the lack of any relation, to these types. 

It has been impracticable to give structural details in the 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 207 

preceding pages and many interesting points must be left 
quite untouched. Some of the major characteristics, how- 
ever, seem to be significant. To these three types belong 
227 genera represented by about 4,923 species. . Among 
these the conspicuous portion of the perianth, calyx or 
corolla, is almost without exception campanulate or more 
generally widely patent and sometimes reflexed. The 
stamens regularly show an elongate form and are basifixed 
on long or short filaments. The pistil is usually simple 
with filiform style and small, punctiform stigma, but to 
this there are rare exceptions. 

The Dilleniaceous type is distinguished from the others 
by numerous stamens and sometimes separate pistils. The 
number of genera assigned to this type is few. In it, de- 
hiscence by pores seems to be the least specialized of these 
three types. The anthers are for the most part elongate 
but rare exceptions to this general rule are to be found ; 
the pores are riot infrequently continued down the sides in 
longitudinal slits. 

In the Solanum-Cassia and the Melastomataceous types 
there is, as compared with the preceding, a reduction of 
the number of the stamens and the pistil is always simple, 
usually with a punctiform but sometimes slightly lobed 
stigma. 

The minor differences exhibited by the individual genera 
and species of the Solanum-Cassia type are somewhat con- 
fusing but the conviction of the existence of an essential 
similarity in the organization of the flowers of the forms 
assigned to this type becomes stronger as the material is 
studied. Here, as everywhere else, nature seems in large 
measure regardless of fixed categories, and yet the struc- 
tural agreement of many of these genera or species from 
widely separated families and of discontinuous geographical 
distribution — an agreement which is often so close that 
one would almost assert that the flowers of the several un- 
related genera indigenous in the most widely separated por- 



208 MISSOURI BOTANICAL GARDEN. 

tions of the globe had been, cast in the same mould or 
formed by the same skilled artisan — is so exact and the 
general structural habit of the aberrant forms is with all of 
the minor differences so uniform in essentials that it seems 
necessary to attribute it to some single factor or co-opera- 
tive group of factors of environment. 

Before we attempt to determine this factor, however, it 
will be well to assure ourselves as fully as possible of the 
real existence of such a structural uniformity. 

Of the structural characteristics, our knowledge is fairly 
satisfactory, although many points can be determined only 
by the examinations of living material. The large series 
of icones, many of them executed from living material, 
in the library to which I have had access while carrying on 
this work has enabled me to make extensive comparisons 
otherwise impossible. It may, I think, be demonstrated 
beyond confutation, that the apically dehiscent forms 
other than the genera which have been assigned to groups 
designated as the Araceous, Gramineous, Polygalaceous 
and Ericaceous types (and all of which are, with the ex- 
ception of the last, perfectly natural and sharply defined 
categories) may also be grouped in classes few in number 
and uniform in their essentials of structure. 

The close structural agreement among themselves of the 
apically dehiscent forms assigned to any class is relatively 
easy of demonstration or refutation although specific varia- 
bility in many genera, the presence ,of minor structural 
characteristics which tend to obscure the more essential 
features, and the inadequacy of our knowledge of many 
forms, renders this no light task. 

The demonstration of a general structural similarity in 
all the parts of flowers, which have been brought together 
for comparison on the basis of a single characteristic — in 
the present case the mode of dehiscence of the anther — 
indicates that there exists a correlation between this 
character and the other structures of the same flower. 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 209 

The nature of this correlation we cannot consider here. 
For the present, we are concerned only with ascertaining 
if an interdependence really exists. An examination of 
the apically dehiscent genera and species alone yields very 
convincing evidence in favor of this hypothesis, but theories 
based upon one class of data are open to criticism and 
especially so when the mass of material is so small as it 
necessarily is in the few apically dehiscent genera assigned 
to the Dilleniaceous or the Solanum-Cassia type. 

ABERRANT NATURE OF APICALLY DEHISCENT FORMS IN 
THEIR SYSTEMATIC GROUPS. 

A satisfactory kind of supplementary evidence will be 
furnished by a comparison of the apically dehiscent genera 
from the several families with the other members of the 
same systematic group. If in addition to an approximate 
conformity to the characteristics of their own type in an 
artificial class limited primarily by a single character, the 
several genera are found to be aberrant in the groups of 
the phylogenetic system in respect to the assemblage of 
their floral characteristics, we shall be justified in conclud- 
ing that there is some direct and demonstrable relation 
between the selected character and the others pertaining 
to the flowers under consideration. 

A thorough comparison such as that here suggested in- 
volves numerous almost insurmountable difficulties. Tax- 
onomists are by no means agreed as to the limitations of 
systematic groups or as to their monophyletic or poly- 
phyletic origin, and under these circumstances what shall 
serve as our basis of comparison? Not only are the limits 
of groups variously and ofttimes ill defined but the data 
available on the floral structure of species or genera are 
sometimes very meager. The labor involved in searching 
through many volumes of descriptions and figures to secure 
the data for such comparisons is very great and even then 

14 



210 MISSOURI BOTANICAL GARDEN. 

the many details cannot be satisfactorily known- Few 
herbaria are rich enough in material to permit of an exami- 
nation of a majority of the forms which should be con- 
sidered and even were this the case the time and material 
required would in most cases be out of all proportion to 
the results obtained. 

In making these comparisons all characters must be 
taken into consideration. The Solanum-Cassia type is not 
distinguished by linear, basifixed, or apically dehiscent 
anthers alone, or by a patent corolla, but by the association 
of all these characters — the patent corolla and the linear, 
basifixed anthers approximated around the simple style and 
opening by terminal pores. Excepting the essential char- 
acter of dehiscence, any one of these may usually be found 
and often represented by a large number of genera in the 
family or group from which the apically dehiscent genera 
under consideration are taken but the combination of all 
the characters except that of the apical opening of the 
anthers in one flower is, in the systematic groups which I 
have examined, much more rare. 

The general results of a quite detailed examination of a 
number of the families, are presented below: — 

Dilleniaceous Type. 

We find in the Dilleniaceae that while the numerous 
structural differences are very confusing it seems quite clear 
that apical dehiscence is confined almost exclusively to those 
forms with long, basifixed anthers. The presence of sev- 
eral free, spreading pistils in this family is unique among 
flowers with apically dehiscent anthers. 

In the Elaeocarpaceae dehiscence may be by pores or 
longitudinal slits in the same genus, all genera but one 
having at least some species with apical pores. This single 
exception is A.niholoma which has the essential floral struc- 
ture of other elaeocarpaceous genera except for the gamo- 
petalous, urceolate corolla instead of the usual polypetalous, 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 211 

generally more or less campanulate, perianth of other repre- 
sentatives of the family. The stamens in this family are 
usually linear and basifixed on filaments of varying but 
usually considerable length and, in this respect, conform to 
the ideal type of this class. 

Elongate, basifixed, usually apically dehiscent anthers 
are characteristic of the Ochnaceae. Other points will be 
taken, up under the Solanum-Cassia type to which the most 
of the representatives of this family have been assigned. 

Of the Theaceae my knowledge is insufficient to permit 
of comparisons. 

JBixa of the Bixaceae is an anomalous form. 

Solanum-Cassia Type. 

Turning now to the Solanum-Cassia type we find mate- 
rial which yields more satisfactory results. 

The Mayacaceae is monotypic and its relationship is not 
so evident as to justify comparisons of its floral structure 
with that of other families. 

All of the genera of the Eapateaceae are apically dehis- 
cent. The family is considered most closely related to the 
Xyridaceae and Eriocaulaceae. 

A patent perianth is characteristic of the Commelinaceae 
and, except for a slight elongation of the anthers, the 
apically dehiscent genera do not differ essentially from 
some others to be found in the family. At the same time 
the various structure of the anthers and the filaments of 
many of the genera distinguish them at once from the 
apically dehiscent forms. 

In thePontederiaceae, Monochoria , the anthers of which 
open only at first by apical pores which are soon continued 
down the sides in slits, has the corolla tube most reduced 
and the limb more widely patent than any forms except 
some species of Heteranthera with a long tube. The fila- 
ments and anthers are of about equal length, but the 



212 MISSOURI BOTANIC Ali GARDEN. 

anthers have probably the greatest relative length in the 
family. 

The floral structure of the Liliaceae and Amaryllidaceae 
is hardly to be summarized in general terms. We may, 
however, repeat that in the Amaryllidaceae the prevalent 
habit of the perianth may perhaps best be described as in- 
fundibuliform, while in the Liiliaceae it is more frequently 
patent or campanulate. In the Amaryllidaceae the four 
apically dehiscent genera stand almost alone as representa- 
tives of the Solanum-Cassia floral habit while in the Liili- 
aceae there are many genera with patent or broadly cam- 
panulate perianth and in some of these the filaments are 
reduced in length, and the anther is of the more elongate 
form commonly seen in the apically dehiscent genera. 

Sollya and O heir anther a are quite aberrant forms in the 
family Pittosporaceae in which they belong, both as 
regards form of perianth and structure of stamens. 

In the Leguminosae the apically dehiscent genera belong 
to a group in which the perianth is generally patent, but 
they differ in the relative length of the filament and anther 
and in the insertion of the anther upon the filament. 

The Treniandraceae contains only apically dehiscent 
genera and species. 

The sterculiaceous genera which have been mentioned 
as having anthers opening at least at first with pores are 
marked by a great similarity of structure, a greater similar- 
ity than that between the apically dehiscent genera and 
the other genera of the Lasiopetaleae, to which all of the 
apically dehiscent genera belong, in which the anthers open 
from the first by longitudinal slits. The Lasiopetaleae 
itself differs very essentially in its floral structure from 
other tribes of the Sterculiaceae so that there can be no 
question as to the unique structure of these forms in this 
family. 

The Ochnaceae has the elongate, basifixed anthers of the 
Solanum-Cassia type and most of the genera have anthers 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 213 

which open at least at first by terminal pores. Since the 
floral habit of this family conforms to that of our Dillenia- 
ceous and Solanum-Cassia types, the only opportunity for 
comparisons is between the genera which may be assigned 
to these two types and certain aberrant forms to be men- 
tioned below. We may postulate that the Solanum-Cassia 
type has in some cases been derived from the Dillenia- 
ceous type by a numerical reduction of the androecium ac- 
companied in most instances by a shortening of the fila- 
ments. On the whole, dehiscence by pores seems to be 
more perfected in the Solanum-Cassia type. A family con- 
taining representatives of both types should furnish data 
of value for a study of this question but, as pointed out 
above, the evidence available in the present case is too 
meager. A strong reduction in the androecium has clearly 
taken place so that the family is broken up into more than 
one floral type, but whether this reduction occurred entirely 
subsequent to the development of the apically dehiscent 
habit or whether it largely preceded it is a question which 
cannot be easily decided. The anthers of some species of 
this family which have been assigned to the Solanum-Cassia 
type show no more specialized apical dehiscence than some 
which have been assigned to the Dilleniaceous type. On 
the other hand it is to be noticed that certain genera are 
very suggestive of the Melastomataceous type — a group of 
forms the evolution of which has apparently progressed 
along a very different line from that of the Solanum-Cassia 
type — and this may explain the equal perfection of apical 
dehiscence in forms with numerous stamens with long fila- 
ments and in those with few stamens with short filaments. 
In certain anomalous genera the staminodia are so modi- 
fied as to form a more or less perfected tubular or urceolate 
corona. In these genera, the anthers open by longitudinal 
slits and since the same condition prevails in other genera 
with an urceolate perianth in families in which the perianth 
is usually patent and the dehiscence apical, it seems legiti- 



214 MISSOURI BOTANICAL GARDEN. 

mate to conclude that the exserted condition is a factor of 
importance in the development or permanence of the apically 
dehiscent anther. Comparative evidence would suggest 
that these forms with cylindrical or urceolate perianth are 
not primitive, but of recent origin. The anthers seem to 
open at first by a terminal gap and it may be that longi- 
tudinal dehiscence has been reassumed by these forms as a 
result of the changes in the form of the perianth. 

Only a thorough knowledge of the living plants in their 
own environment supplemented by histological studies will 
decide some of these questions. 

Stemonoporns and ]$£onoporandra differ in the structure 
of the androecium from all other Dipterocarpaceae. 

In Begonia the perianth is patent in all forms. The 
two apically dehiscent species of the section Solanthera 
have more elongate anthers than many species of the genus, 
but the distinction here is not a very sharp one. 

The structure of Gardrieria seems to be unique among 
the Loganiaceae. 

In the Grentianaceae a few forms approach very closely 
the structure of the apically dehiscent genera and many 
agree in some characteristics, but on the whole it seems to 
me that JExacum and Ootylanthera depart markedly from 
the type of the family. 

A long corolla tube is characteristic of most Solanaceae. 
Several genera have a more or less widely open corolla. 
Among these, Solarium is the genus in which the anthers 
have attained the greatest length and the filaments are most 
reduced. 

Comparisons in the Rubiaceae are rendered very difficult 
by the size of the family, the inadequacy of our knowl- 
edge of the floral structure and especially by the large 
number of genera which show some of the characteristics 
of the flowers of the Solanum-Cassia type ; but when all 
characters are considered, Argostem/ma, Neurocalyx and 
Strumpfia seem to be unique in this vast family. 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 215 

Melastomataceous Type . 

In the Melastomataceous type, the Melastomataceae need 
not engage our attention here. "We have already noticed 
the similarity of floral structure in the Melastomataceae. 
The floral structure is quite unique and a comparison with 
groups to which it is more or less closely related is not 
possible in this place. 

The essential structural difference between the Solanum- 
Cassia type and the Melastomataceous type seems to be the 
length of the filaments. The long filaments make possible 
the highly organized anthers found in the latter. The 
structure of the flowers need not be redescribed here. The 
essential similarities may be gathered from any series of 
descriptions or plates of the genera of this family, while the 
minor details are too numerous and perplexing for consid- 
eration. 

The species from other families which have been as- 
signed to this type are very few. Storckiella is the only 
genus of the Cassieae with elongate filaments. JSfaximil- 
iania and Amoreuxia of the Bixaceae have the form of 
anther which is so frequently seen in this type and long, 
filiform filaments. The stamens are more numerous than 
is usually seen in the Melastomataceae. The Bixaceae 
comprises four genera belonging to three tribes. The first 
contains the unique Bixa, the second Maximiliania and 
Amoreuxia and the third SpJiaerosepalum, exhibiting nu- 
merous stamens with long, filiform filaments and nearly 
globose, dorsifixed, versatile anthers dehiscing by longitu- 
dinal slits. The structure of the two genera which I 
have assigned to this type is, then, quite unique in this 
family. 

As I have insisted above, the limitation between the 
Solanum-Cassia and the Melastomataceous type is not a 
sharp one and it may be that such forms as Cheiranthera 
of the Pittosporaceae and Exacum of the Gentianaceae 
would have been best treated under this type. Here, a 



216 MISSOURI BOTANICAL GARDEN. 

knowledge of the living plants such as may be gathered by 
those who know them in the field is most desirable. 

RECAPITULATION . 

The evidence assembled in the preceding paragraphs 
seems to point clearly to certain easily stated conclusions. 
Not only do the apically dehiscent forms show a high de- 
gree of similarity inter se but the several forms are very 
frequently aberrant in respect to the totality of their 
characters in the systematic groups to which they belong. 
In other words, apical dehiscence of the three types here 
considered occurs only in flowers of a certain structural 
habit. The objection which might be raised that I have 
included in these three groups only such forms as are 
similar in the totality of their characters is disarmed by 
the fact that these three classes are sharply distinguished 
from the first four types recognized and that anomalous 
forms are very few. The fidelity to type of the apically 
dehiscent genera and in many cases their aberrant floral 
habit when compared with other members of the same 
systematic group speak strongly in favor of a correlation 
between the floral parts and the form and mode of dehis- 
cence of the anther. The exact significance to be assigned 
to the broad term correlation must be defined in each case 
if ambiguity is to be avoided. Interdependence of parts 
may be real or apparent, due to internal or external forces. 
In this paper, I have sought to avoid all theoretical con- 
siderations and I shall not discuss the nature of the corre- 
lation observed in these forms. I have used the term in 
the broadest sense, indicating any reciprocal relationship of 
parts ; the ground for the assumption of such a relation- 
ship is found in the constancy with which the characteris- 
tics in question are found associated in the apically 
dehiscent forms. 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 217 

Floral Ecology or Apically Dehiscent Forms. 

It was the similarity of adaptation which first aroused 
interest in the morphological peculiarities described above. 
We may now examine the data available upon these floral 
structures regarded as adaptations. 

The ecology of the Araceous, Gramineous and Poly- 
galaceous types is quite foreign to our present considera- 
tions. That of the Ericaceous type has many points of 
interest but material is not yet satisfactorily arranged for 
a summarized statement and the individual data are too 
numerous for consideration here. 

ECOLOGY OE DILLENIACEOUS TYPE. 

In the Dilleniaceae, Ducke saw Melipona bipunctata 
abundant On an unidentified Tetracera at Para in Brazil 
and also a small species of Halictus visiting the flowers of 
Davilla rugosa in great numbers. On Saurauia the 
interesting habit of producing inflorescences from the lower 
regions of the main stem has been described in some detail. 
Knuth observed in Java numerous small insects upon the 
flowers of S. cauliflora: *' Vor allem war Thrips sehr haufig 
in den Bliiten, Musciden flogen unstet von Bliite zu Bliite 
und hielten sich in jeder einzelnen mehrere Sekunden auf , 
mit grossen Ballen weissen Pollens wieder hervorkommend, 
und auch kleine Bienen (PodaUrius?) flogen ab und zu." 
As a visitor of $. nudifiora he saw JPodalirius clinging to 
the flowers from below and diligently collecting pollen. It 
is unfortunate that the evidence is so meager that one is 
only able to surmise that these are pollen flowers, perhaps 
very similar to Rosa. 

Observations on the Elaeocarpaceae are of the character- 
istic meagerness. According to Thomson, Elaeocarpus 
Hookerianus of New Zealand has perfect, white, protan- 
drous flowers in which a glandular ring at the base of the 
stamens secretes nectar freely ; it appears to be entomophil- 



218 MISSOURI BOTANICAL GARDEN. 

ous. Dahl saw the honey less (?) flowers of E . Panhinsoni 
visited by two birds — Oharmosyna and JSfyzomela [for 
what purpose?]. In Aristotelia Jifaqui there are said to be 
bright yellow male flowers with two whorls of stamens and 
smaller physiologically female flowers with one whorl. In 
Chile Johow found the flowers industriously visited by 
the honey bee. 

In New Zealand according to Thomson A., racemosa 
shows transition stages from hermaphrodite to purely 
staminate or pistillate flowers. These are red and desti- 
tute of odor or nectar ; the light, pulverulent pollen sug- 
gests anemophily. A., fruticosa of the same region has 
polygamous flowers which lack odor and nectar, but here 
anemophily is doubtful. 

The unique form of the anthers and perianth in the 
Elaeocarpaceae renders a fuller knowledge of their floral 
ecology highly desirable. 

On the large, red flowers of Bixa, of the Bixaceae, 
Ducke observed at Para the females of several species of 
bees belonging to Oentris, Euglossa, Bombus, Xylocopa, 
JSfelipona, especially large species, and Halictus. 

ECOLOGY OE SOLANUM-CASSIA TYPE . 

Fortunately our knowledge of the floral ecology of the 
members of the Solanum-Cassia type is much fuller and 
so it will not be so necessary to cite the special observa- 
tions as it has for the Dilleniaceous type. 

Delpino divides the twelfth class (Apparecchi prensili) 
of his biological classification of floral forms into two types, 
" Tipo Boragineo " and " Tipo Verbascino." The first 
of these is of very particular interest in this place since 
the Solanum-Cassia type as limited in this paper is, so far 
as observations and conclusions from analogy go, very 
nearly synonymous with it. 

The Borago floral type is, according to Delpino, charac- 
terized by pendulous or quasi-pendulous flowers. The long 



THE DEHISCENCE OF ANTHEKS BY APICAL PORES. 219 

anthers, inserted on short, thick filaments, are connivent 
in a cone, through the center of which passes the style. 
Dehiscence is at the apex by pores which may be continued 
longitudinally. The pollen is dry and smooth, and neces- 
sarily falls upon the ventral surface of the visiting insect 
at the moment when it grasps the staminal pyramid. In 
order that the insect may cling to the staminal column 
the expansion of the flower is generally considerable. 
Nectar is absent or present only in small quantities. In 
the former case the bee collects only pollen. This type 
is exclusively melittophilous, "e sorprendentemente si 
ripete co' suoi essenziali caratteri in molte famiglie di 
piante, cioe' nelle Boraginee, Primulacee, Solanacee, 
8crofulariacee, jLmarillidee , Asparaginee , Pitlosporee, 
-Loasacee." 

In this type he places, with comments upon many forms, 
Borago officinalis, Cyclamen europaeum, C. count, O.per- 
sicum and other species, DodecatJieon Jifeaclia, D. integri- 
folium, Solanum Dulcamara, 8. nigrum, 8. tuberosum, 
8. Lycopersicum, 8. insanum and many other species of 
the genus, Verbascum Myconi, Q-alanihus nivalis, Leuco- 
jum vernum, Conanthera bifolia, CajopJiora lateritia and 
many species of Loasa, 8ollya linearis, Dianella coerulea 
and other species of this genus. 

Delpino's division of floral forms into fixed categories 
has been censured, and one of the most severe criticisms 
by Miiller refers to this type. Without intending an eval- 
uation of the classification as a whole I must say that it 
seems to me that Delpino's generalizations for this type 
have been very satisfactorily substantiated by the obser- 
vations which have been accumulated since his essay ap- 
peared. In their relations to insects the forms which are 
here placed in the Solanum-Cassia type are in close agree- 
ment with Delpino's Borago type. For the purposes of 
this paper the classification is extended so as to include the 
zygomorphic forms so common in Cassia and occurring in 



220 MISSOURI BOTANICAL GARDEN. 

some other genera, forms which should perhaps be placed 
in the Melastomataceous type as Delpino has done. It 
must be recognized, however, that between these types 
there is no sharp distinction and that the groupings 
adopted are merely for the purpose of more convenient 
comparisons. In treating the floral ecology of the Sola- 
num-Cassia type it seems best to discuss in some detail the 
two type genera and compare with them other members of 
their respective families, taking up afterwards the other 
forms in their systematic sequence. 

We may first consider briefly the actinomorphic Sola- 
nums. 

8. Dulcamara has been most thoroughly studied. It 
belongs clearly to Delpino' s Borago type. The Syrphidae 
and Lepidoptera observed as visitors by Miiller must 
be regarded as accidental or insignificant. According to 
most reports, 8. Dulcamara is little visited, but Hoffer 
found the visitors in one locality very abundant and 
active, no less than seven species of Bombus and one 
of Osmia, as well as ffliingia, Volucella and Argynnis 
being observed. 8. nigrum and 8. Oarolinense have 
been observed to be sparingly visited by Bombus. On 
8. elaeagnifolium at Las Vegas, N. M., Cockerell took 
eighteen species of Apidae belonging to fifteen different 
genera. At Para, Ducke observed as visitors of 8. grandi- 
florum (said in systematic works to have a violet corolla 
2£— 3 inches in diameter) one species each of Oxaea, Xylo- 
copa, Bombus and Halictus and six of Oentris. At the 
same place he observed one species of Halictus, seven 
species of Buglossa and a large species of Melipona visit- 
ing the flowers of 8. toxicarum. On an unidentified species 
with blue flowers he records only species of Melipona, 
especially JS£. fasciata, Bombus cayennensis and Halictus. 

Darwin reports observations by F. Miiller on 8. pollina- 
canthum at Sta. Catharina, Brazil. He found it visited ex- 



THE DEHISCENCE OF ANTHEES BY APICAL PORES. 221 

clusively by pollen collecting bees, 3felipona, Euglossa, 
A.ugochlora, Megacilissa, Eophila and others. 

I have observed that 8. sisymbrifolium has a very pro- 
nounced perfume, suggesting more abundant insect visits 
than in many other species in which no fragrance is to be 
detected. In the summer of 1903, 1 found large bees with 
the greatest frequency collecting pollen. In his paper on 
Brazilian solitary bees Schrottky gives 8. Balbisii (== 8 . 
sisymbrifolixiTti) as one of the principal flowers visited by 
JSylocopa . 

Though few, these observations indicate clearly the true 
method of pollination in the actinomorphic species of this 
large genus. A few species are zygomorphic. Upon two 
of these, direct observations have been made. Both have 
been shown to be exclusively adapted to pollination by the 
larger bees, A.pis and Bombus having been observed as 
well as some of the smaller Apidae. The numerous inter- 
esting points incident to their zygomorphy and their ac- 
companying lateral asymmetry cannot be detailed here. 

Between the zygomorphic Solanums and Cassias a most 
remarkable ecological similarity prevails and it was this 
which first called attention to the problem in hand. 

The floral ecology of the genus Cassia has attracted con- 
siderable attention, the first suggestive observations being 
those of Delpino and Leggett in 1871 to 1881. The paper 
by Todd on 8olanum rostratum and Cassia chamaecrista 
called particular attention to both of these genera and has 
been the incentive to most of the later work. The more 
important papers are those by Todd, Miiller, Burck, Harris 
and Kuchs and the recent treatment in the third volume of 
Knuth's Handbuch. 

Since the appearance of Todd's paper and the more gen- 
eral one by Miiller the attention of writers has been chiefly 
devoted to the " division of labor " in the stamens and the 
curious phenomenon of right- and left-handedness, or enan- 
tiostyly as it has been called. 



222 MISSOURI BOTANICAL, GARDEN. 

These morphological features of Cassia may not be dis- 
cussed here although they have figured conspicuously in 
all the ecological discussions pertaining to these genera. 
Their true significance has probably not yet been correctly 
interpreted. 

On O. cliamaecrista in Iowa Todd saw a small humble 
bee. In Illinois Robertson observed four species of Bom- 
bus collecting pollen. *' Landing upon the anthers, they 
seize them between their mandibles and stroke them down- 
wards with a sort of milking motion." In a later paper 
he states that the flowers are visited by the pollen collecting 
female of the oligotropic JPodalirius walshii. In eastern 
Kansas, Harris and Kuchs observed one species each of 
Jlpis, jAgapostemon, JMelissodes and Bombus. Near St. 
Louis I have observed Bombus. 

According to the observations made by Meehan, Robert- 
son and myself, '. JSfarylandica is visited by Bombus. 
Pollen is collected only from the smaller upper stamens, 
the three lower ones serving the insect only as a sup- 
porting platform if at all. Flowers protected against 
insect visits do not produce fruits. The pollen from the 
large lower or the three smaller upper stamens, may effect 
fertilization, but whether there is a difference in the effi- 
ciency of that from the two kinds of stamens is not 
known. 

In an unidentified species from Brazil related to O. 
laevigata .Fritz Miiller observed Bombus violaceus and 
species of Oentris collecting pollen from the short, central 
anthers while the two long, lower anthers and similarly 
shaped pistil are so much curved as to strike with the 
anther tips and the stigma the dorsal surface of the 
abdomen of the visiting insect. The large, lower stamens 
are of the same color as the smaller, central ones and 
were observed to be visited extensively by small insects 
which despoiled them of their contents without effecting 
pollination : Trigona elegantula, T. lilliput and a large 



THE DEHISCENCE OP ANTHERS BY APICAL PORES. 223 

species of Augochlora. Trigona ruficrus often bites the 
anther quite in pieces. 

Burck considers that the floral mechanism of Cassia 
favors autogamy, basing his conclusions on the fact that in 
several species studied by himself the tip of the pistil 
returns to the median plane of the flower by a new curva- 
ture of the style so that it will not come in contact with 
the region of the insect's body previously supplied with 
pollen from the pores of the larger stamens, and that in 
others the stigma comes almost in contact with the tips of 
the stamens. 

Burck' s paper contains some interesting observations, 
but it seems that his conclusions can hardly be accepted 
without further investigation. For a more detailed com- 
parison of his statements and conclusions, reference must 
be made to his own paper. 

According to him, C '. glauca differs from the other 
species in having coi"iaceous instead of indurated anther 
walls. The large bees — Xylocopa and Bofnbus — which 
visited this, and other species observed by him, alight in 
the middle of the flower and ' ' devour the grains of pollen ' ' 
of the five short stamens or they may destroy in part or 
entirely both the anther walls and the pollen. Only the 
two large stamens remain intact and undespoiled of their 
pollen. Knuth's observations on O. Horsfeldii, considered 
by some a synonym of C. glauca, confirm Burck' s obser- 
vations on this point. The bee flies directly towards the 
small stamens, grazing but rarely the large anthers; when 
these are touched, it is with the ventral surface of the body. 
He never saw the stigma in contact with the body of the 
visiting insect but thinks that this might occur, especially 
at the moment when the insect leaves the flower. But 
even then, the chances are that the dorsal surface of the 
insect's body, rather than the ventral surface to which the 
pollen adheres, will touch the stigma. 

The significance of the elongation of these lower anthers 



224 MISSOURI BOTANICAL GARDEN. 

is, he thinks, their protection from the insect visitors. 
According to the observations of O. Schmiedenknecht this 
species in the garden at Buitenzorg is visited principally 
by species of Oeratina. O. bacilluris has indurated 
anthers, so that, according to Burck, the insect can secure 
pollen only by inserting its proboscis into the terminal 
pores. Knuth speaks of the anthers being so completely 
milked out that only a few grains remained. In the 
Buitenzorg garden, 4 species of Xylocopa and unidentified 
Syrphidae were observed as visitors. 

O. alata has been described by Burck, Landman, Ducke, 
and Knuth. Lindman saw large and small bees visiting 
this form in large numbers. He states that in this species, 
as also in O. aculeata, the large anthers sometimes con- 
tained no pollen even in the bud. Ducke saw at Para and 
Macapa, 4 species of Oentris, 3 of Xylocopa and 4 of 
Euglossa. At Buitenzorg, Oeratina was the principal 
visitor observed. 

Some of the species studied by Knuth at Buitenzorg have 
already been mentioned. We may summarize his observa- 
tions on the other species by saying that he usually found 
only Xylocopa as a visitor. In rare cases he observed 
Podalirius zonatus and ^Megachile opposita. Oentris and 
Euglossa do not occur there. O. Schmiedenknecht also 
observed several species of Oeratina on Cassia flowers at 
Buitenzorg. Knuth thinks that cross fertilization is favored 
since the projecting stigma will be touched first by a visit- 
ing insect. In all cases supplementary self-pollination may 
occur when the flowers are visited by bees, but spontane- 
ous autogamy he considers possible only in the species with 
small flowers which close after a time, bringing the stigma 
into immediate contact with the anthers. In St. Louis, I 
have taken Bombus and smaller bees collecting pollen from 
O. sophera. 

According to Miiller, O. vnultijuja at Blumenau, South 
Brazil, is visited by Xylocopa artifex and another large, 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 225 

apparently undeseribed species, and by Centris lineo- 
latus. 

Trelease saw C. obtusifolia visited by Bombus in central 
Alabama. 

Schrottky observed Centris discolor regularly collecting 
pollen on C. splendida at S. Paulo, Brazil. Here he also 
saw species of Centris collecting the pollen of C. bicap- 
sularis. 

AtPara, Ducke observed on C. Hoffmansegii and other 
species with very large flowers, Xylocopa, Centris, and 
JSuglossa. 

In Chile, Johow saw C '. closiana visited by Bombus 
chilensis. 

In Chile, Darwin saw a Mimus with its head yellow with 
■pollen, from, as he thought, a Cassia. Delpino records 
Cassia (?) as visited by JVectarinia platura. Fries once 
■observed C. bicapsularis visited for insects by Chloros- 
iilbon auriventris. He found the species regularly visited 
and pollinated by Bombus carbonarius. 

Our knowledge of forms other than Solanum and Cassia, 
is, so far as direct observations are concerned, very frag- 
mentary. Brown studied the flowers of Labichea lan- 
ceolata, and, finding the same right- and left-handed ar- 
X'angement as in S. rostratum, concluded that the method 
of pollination is the same. It can hardly be doubted that 
the other forms of the Cassieae are, at least for the most 
part, dependent upon the larger pollen-collecting bees for 
their fertilization. 

On the ecology of the Mayacaceae and Rapateaceae no 
observations have been published. 

Of the floral ecology of the Commelinaceae under con- 
sideration we are quite ignorant. I have been able to ex- 
amine living material of one species of Dichorisandra and 
to convince myself that there is no secretion of nectar. 
The ecology of the flowers of some genera of the family is 
quite well known, and comparative evidence makes it very 

15 



226 MISSOURI BOTANICAL GARDEN. 

probable that most of the Commelinaceae are without nec- 
tar and are visited by pollen-collecting bees. 

Before leaving this subject, I will refer briefly to the 
anomalous Cochleostema. It is the most magnificent mem- 
ber of the Commelinaceae, with flowers two or two and 
one half inches in diameter, sepals and petals highly 
colored and patent, androecium of six elements in two 
whorls, fertile stamens 3, 1 from outer and 2 from inner 
whorl, forming a pedicelled staminal column with the 
anthers all spirally twisted and dehiscing longitudinally ; 
one horizontal at the base of a corniform hood formed by 
involute, petaloid outgrowths from the other two which 
inclose all three stamens. 

The habit of this most complex flower is almost exactly 
identical with that of some of the zygomorphic, apically 
dehiscent forms, as for instance, S. rostratum, the form 
of the perianth and the position of the pistil being the 
same and the staminal hood simulating almost exactly 
the arcuate anther. Loew and Appel in Knuth's Hand- 
buch devote considerable space to a discussion of this 
remarkable plant and I shall add the further suggestion 
that the fragrant, nectarless flowers may be adapted to the 
larger Apidae which seek the hairs of the staminodia or 
other juicy portions of the flower and effect pollination in 
the same way as in apically dehiscent forms. 

When I first saw a habit figure of Oochliostema I thought 
it an apically dehiscent form with one large anther similar 
to S. rostratum. If the suggestion just offered prove the 
right one, we have a most remarkable case of parallel 
adaptation. 

Nothing is known of the pollination of JSIonochoria. 
JPontederia is visited for nectar and sometimes for pollen 
by large bees. Heteranthera has been described as a pollen 
flower. The secretion of nectar is not abundant if it occurs 
at all in Eichomia, at least when grown in the north. On 
IS. azurea, Miiller observed only bees belonging to the 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 227 

genera Melipona and A.nthophora. I arn inclined to be- 
lieve that in JSfonochoria we have a highly developed pollen 
flower, connected by various stages with other members of 
the family showing a deep nectar bearing tube. 

In the Liliaceae, all forms of which, with the exception 
of the aberrant Odontostomum , belong to the Solanum- 
Cassia type, zygomorphy is not so pronounced as in several 
other families, the Amaryllidaceae for instance, appar- 
ently only A.grostocrinum having a zygomorphic habit. 
The flowers are all of a good size and usually highly colored, 
the anthers, quite often in sharp contrast with the perianth. 
Thomson states that Luzuriaga parviflora has white, 
fragrant pollen flowers ! Delpino considers Dianella coeru- 
lea and other species as pollinated in the same way as Borago 
or Solarium. 

In the Amaryllidaceae, direct observations are wanting, 
but the zygomorphic condition of Zejphyra and Tecophilea 
and especially of Oyartella indicate ecological relations 
similar to those of the zygomorphic Solanums or Cassias. 

Nothing is known of the ecology of the two representa- 
tives of the Pittosporaceae considered, but their structure 
is so similar to species which are well studied biologically 
that there can be little doubt as to their insect visitors. 
The little that is known of other genera indicates a wide 
difference in the insect relations of the typical and the aber- 
rant members of the family. 

The Tremandraceae are doubtless entirely dependent 
upon pollen-collecting bees for cross fertilization. This 
cannot be said with the same confidence of the Sterculiaceae 
considered above, but from the structure of some of the 
forms it seems not improbable that they are visited and 
pollinated in the same manner as Solanum or Borago. 

Gilg considers that in the absence of direct observation 
it is to be assumed without question that the flowers of the 
Ochnaceae are insect pollinated. The magnificent flowers, 
the zygomorphic disposition of the anthers in some species, 



228 MISSOURI BOTANICAL. GARDEN. 

especially the peculiar condition of the anthers connate in 
masses in Luxembergia, and the pleasant odor observed for 
others (the odor of violets in Schuurmansia Henningsii) 
all speak strongly in favor of this conclusion. Visitors are 
recorded for but one species. Ducke observed Xylocopa 
barbata visiting the flowers of an unidentified Ouratea. 

Of the method of pollination of the Dipterocarpaceae we 
know nothing, and can only surmise from the structure of 
the flowers. The same is true of the Flacourtiaceae under 
consi deration . 

Nectar has never been demonstrated in the showy but 
usually odorless flowers of Begonia. "Warburg noticed 
numerous small insects visiting the flowers in their native 
habitat. On an undetermined Javan species Knuth ob- 
served Apis, Sombus and Podalirius (presumably col- 
lecting pollen. ) 

Of the Myrsinaceae and Loganiaceae in question we 
know nothing. 

Mr. J. C. Willis, Director of the Royal Botanic Gar- 
dens, Peradeniya, kindly sends me the following note v on 
Exacum of the Gentianaceae : ' ' The only Melastomaceae 
I have actually seen pollinated yet are our common big 
pink Osbeckias, which are visited by the big carpenter bee, 
JZylocopa, sp., apparently for pollen. The bee pollinates 
our big blue Exacum macranthum, which looks just like 
a Melastomacea. It squats down on the flowers and 
bunches up the stamens with its legs, and must get lots 
of pollen jerked out of the apical pores onto its body . 
In fact I have usually found flowers visited by it to have 
the stigmas well pollinated, and have hardly ever seen any 
other insect visit them at all. There is little or no honey, 
so far as I have noticed, in these flowers or in those of 
Melastomaceae . ' ' 

Solarium has been discussed above. I have been able 
to examine one species of Oyphomandra in a living condi- 
tion. There appears to be no free nectar. The walls of 



THE DEHISCENCE OF ANTHERS BT APICAL PORES. 229 

the anthers are thin and elastic so that the pollen is ejected 
by a bellows-like action as in some Melastomataceae. 
Of the Rubiaceae we know nothing. 

ECOLOGY OF MELASTOMATACEOTJS TYPE. 

As one glances over the icones of the numerous genera 
he is impressed with the high degree of organization and 
the splendid colors, often heightened by the marked con- 
trast between the androecium and the perianth, of the 
flowers of the Melastomataceae. The natural conclusion is 
that these represent forms dependent in high degree upon 
insect visitors. It is very disappointing that our knowl- 
edge of the floral ecology of these forms is so meager. 

Ule remarks: "In vielen Fallen scheint es bei den 
schonbliihenden Melastomaceen nur darauf anzukommen, 
dass die Staubgefasse in Bewegung gesetzt werden, und 
deshalb stehen sie auch noch auf einer niederen Stufe der 
Anpassung an Insektenbesuch . Der Mangel an Insekten, 
welche die Bliithen der Melastomaceen aufsuchen, verur- 
sachte wahrscheinlich das intensive Farbenkleid mit dem 
viele die Gegenden zu -Zeiten schmiicken, wie es keine 
andere Pflanzenfamilie in Brazilien thut. Ich erinnere hier 
an andere auffallende und grosse Pollenblumen, wie z. B. 
Rapaver, Tulipa, Rosa, etc." Ducke says of the family 
at Para: ** Es ist auffallend, dass diese hier so artenreiche 
Familie nur sehr wenig von Apiden besucht wird ; hoch- 
stens findet man ab und zu Halictus an diesen Bliiten. 
Nur eine baumartige Melastomacee sah ich hier ofters von 
Hummeln (JBombus cayenensis~) umschwarmt." 

Leggett first described the bellows-like action of the 
stamens of Rhexia Virginica and observed Bombus visiting 
the flowers. Unfortunately he does not state whether 
nectar is present. According to Bailey the pollen of Hete- 
rocentron roseum is ejected in the same way. According 
to Miiller and Forbes certain species of Afelasloma and 
Heeria agree in the essentials of their floral mechanism. 



230 MISSOURI BOTANICAL GARDEN. 

The flowers need not be described here in greater detail 
than to state that the larger lower anthers are of the same 
color — violet or red — as the vertical patent corolla, 
against which the shorter, bright yellow, central stamens 
form a striking contrast. Miiller observed only a small fly 
(Syrphidae) collecting pollen from the yellow anthers 
and Trigona ruficrus visiting first the small anthers 
and then often the larger ones, offering a more gen- 
erous pollen store, as well, working on them with their 
mandibles and often devouring them entirely. Forbes 
speaks of the pollination of such forms as follows : "I 
have witnessed in many instances the visitation by various 
species of large Hymenoptera, such as Xylocopa and 
JBombus, of species especially of the genus JSIelastoma , pos- 
sessing stamens in all points corresponding to that occur- 
ring in the Heeria described in the letter referred to, and 
what takes place seems to be as follows. The large bees 
evidently make for the yellow platform offered by the short 
stamens, perhaps because they do not perceive the pistil 
and long stamens owing to their projection against the 
broad petaled corolla of the same color in the background , 
and invariably receive the pistil between their legs, their 
feet settling on the fork of the connective, the instant 
effect of which is to collect the whole of the long stamens 
into a bunch, and to depress their anthers downwards and 
away from the body of the visiting bee, while the pistil 
remains in constant contact with its ventral side. At the 
moment of the bee's departure the hooks on the bee's feet 
by pulling on the connective fork, raise the anthers of the 
long stamens, so as to bring the tips of the collected bunch 
into contact with its sides and abdomen. Dr. Miiller's 
statement 'by moving the connective fork of the larger 
ones ' is somewhat ambiguous ; for it is movement only in 
one direction that is of avail in raising the anthers of the 
larger stamens, pressure at the connective hook of course 
tends to depress the anthers and keep them apart from the 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 231 

bee's abdomen while a very slight backward pull at once 
raises the anther." 

Forbes and Burck were able to detect differences in the 
pollen of the two types of stamens. " The pollen from 
the short stamens was large and three-cornered, while that 
of the longer ones was very much smaller and of a more 
oval shape ; and while both forms were found on the pistil, 
only the pollen from the long stamens seemed to be fertile. 
We could not detect any of the short-stamened pollen with 
tubes opened." In a later paper Forbes records Bombus 
senex as a visitor. In his paper on division of labor in the 
stamens of pollen flowers, Miiller records further observa- 
tions by his brother in which incipient stages of the color 
dimorphism were observed in a melastomataceous genus, 
probably Pleroma. The significance of color dimorphism 
has already been alluded to under the Solanum-Cassia 
type. 

Ule studied some species of Tibouchina for comparison 
with JPurpurella. The visitors of these flowers with splen- 
did colors he found relatively few while they were much 
more active on the inconspicuous flowers of Oroton, Sper- 
macoce, Sida, etc., growing near by. Butterflies, flies 
and birds do not visit these forms which secrete no free 
nectar but offer only pollen difficult of collection which 
various pollen-collecting Hymenoptera are able to gather. 
The pollen is not so dry as in PurpurelJa but a little 
sticky. A few times he saw humble-bees visiting the 
flowers, and once he was able to examine in detail the 
action of one on a shrub of T. JMbricandiana, " die immer 
schnell den Kopf noch dem Mittelpunkt der Bliithen steckte 
und bald einen Strauch abgeernt hatte. Warscheinlich 
f angt sie so den ausspritzenden Bluthenstaub in den Haaren 
auf und kann dann leicht Fremdbefruchtung erzeugen. 
Haufiger sind Heine und winzige Bienen, welche die diinne, 
gefaltete Antherenhaut durchnagen und so den Pollen ein- 
sammeln, wie ich mich ofter iiberzeugt und sogar die 



232 MISSOURI BOTANICAL GARDEN. 

Bienen mit ihrem pollenbeladenen Hoschen eingefangen 
habe. Bei dieser Arbeit verweilen diese Bienen aller dings 
langereZeit in jeder Bliithe." Another group of insects 
attack tbe stamens and sometimes devour them entirely, 
and he often found large wasps on the flowers of T. 
gloriosa, while in this class are also to be placed Coleoptera, 
as species of Cytonia and Bupreslis which are especially to 
be sought in the handsome flowers of the Melastomataceae. 
The attractiveness of the stamens for insects and the secre- 
tion of nectar by these organs in Purpurella led to a 
microscopic examination of these parts of T. gloriosa with 
the demonstration of an abundance of sugar which, how- 
ever, is never excreted in visible form. In many species 
of Tibouchina a pronounced change in color in different 
ages of the flower is demonstrable, and of this several 
examples are given. 

A note from Mr. "Willis on the pollination of the Melas- 
tomataceae and JZxacum has been quoted above under the 
Solanum- Cassia type. 

Prom the detailed discussion by Lagerheim, Brachyotum 
ledifolium seems to be adapted to pollination by birds. 
The depth of the nectar and the size of the opening in 
connection with the consistency of the perianth and the 
position of the flower tend to exclude insect pollinators 
with the exception of the Sphingidae, and these are hardly 
to be considered on account of the lack of perfume in the 
flowers and the insufBency of a soft proboscis for the 
ejection of the pollen from the anthers. As visitors he 
observed Rhamphomicron herrani and ATelattura lyrian- 
tliina. These inserted their beaks into the small opening 
of the flower for the purpose of obtaining insects or nectar 
secreted by the filaments and forced a puff of pollen from 
the bellows-like anthers by pressing them with the tip of 
the beak. By pressing the elastic anther wall with a 
rounded match the pollen may be ejected for a distance of 
three centimeters, after which the wall assumes its original 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 233 

position and the pollen collects in the small portion of the 
anther until by repeated compressions it shall all have been 
ejected. The stigma projects bejond the anthers and so is 
first touched by the visiting bird. Presumably other 
species of the genus are also ornithophilous, but B. Ben- 
thamianum would seem from the floral structure to be 
dependent upon bees for pollination. 

Very interesting observations were made by Burck upon 
JSfemecylon edule var. ramiflorum . He describes the prom- 
inent, active, yellow nectary on the dark violet connective 
spur. He finds the flowers well protected from a large 
destructive ant by smaller ones which secure nectar secreted 
from the outside of the calyx tube, but not from the stam- 
inal glands. As pollinators he found small, active flies 
such as are concerned in the pollination of Aristolochia. 
These flies do not seem to fear the ants and he suggests 
that after the flower had developed in adaptation to bee 
visitors, these found the more recently arriving ants uncon- 
genial, and upon their desertion of the plant another adap- 
tation to insects of a very different type became necessary 
and then the recent staminal nectaries arose. 

As pointed out in the descriptive portion of this paper, 
the disposition of the petals in the Melastomataceae is 
very rarely campanulate. One such form, Brachyotumi, 
has been described above. In JPurpurella cleisiopetala 
from the Serra do Itatiaia of Brazil, as described by Ule, 
the flowers never open. The discussion of structural 
points need not be repeated here. He at first suspected 
that Bombus was the insect concerned in pollination, but 
repeated observations revealed only a small ant covered 
with pollen. A larger species of ant destroys the flowers. 
Other forms found in the mountains, as _P. hospita, have 
pendulous campanulate flowers. 

According to Darwin Oentradenia fioribunda, Bhexia 
glandulosa, Jkfonochaetum ensiferum, BTeterocentron J\£exi- 
canum and species of Pleroma are self sterile or but rarely 



234 MISSOURI BOTANICAL, GARDEN. 

produce fruit when left to themselves, but produce seed 
abundantly when artificially self fertilized. 

Little is known of forms outside the type family. Ducke 
observed Oentris and Sylocopa visiting the magnificent 
golden yellow flowers of C ' ocJilospermum (=]lfaximilianeay 
insigne but nothing is recorded concerning the objects of 
the visits. 

RECAPITULATION . 

Summarzing this section, we may say that our knowl- 
edge of the ecology of these forms is far from satisfactory. 
Our acquaintance with the ecological relations of the Dil- 
leniaceous type justify no conclusions, but for the Solanum- 
Cassia and the Melastomataceous types nearly all the 
direct observations which we have and the entire mass of 
comparative data indicate adaptation to the Apidae and 
among these, for the most part, to those which are active 
in the collection of pollen. The absence of nectar prac- 
tically limits the range of visitors to pollen-feeding flies 
and pollen-collecting Apidae. The evidence for adaptation 
to the latter is very great. As to the origin or method of 
evolution of these types of adaptation, our evidence will 
not justify any conclusions, but of the reality of their 
existence, we may have a high degree of confidence. 

We must now leave this phase of the subject which I 
hope to take up briefly again after considering the data 
of two other sides of the problem. 

The Geographical Distribution of Apicallt Dehiscent 

Forms. 

In the introduction to this paper, it was suggested that 
the flora of certain regions is richer in apically dehiscent 
forms than is that of others. The examination of this 
hypothesis was one of the principal objects of the investi- 
gations described. 

The Solanum-Cassia type seems to be the one best adapted 
for a first study of this question. 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 235 

Of the Monocotyledons one family, the Rapateaceae of 6 
genera, is endemic in tropical South America. IMayaca 
has 8 of its 10 species there. Of the two apically dehis- 
cent Commelinaceae, the genus in which dehiscence by 
pores is mostperfect is endemic in tropical South America, 
as is also the highly interesting Oochliostema. In the 
Liliaceae and Amaryllidaceae, 4 of the 8 apically dehiscent 
genera are endemic in extratropical South America. 

Thus, 11 of the 19 genera of the Monocotyledons are 
endemic in South America and another has 8 of its 10 
species there. 

Turning now to the Dicotyledons, we find that of the 40 
genera, 11 are endemic in South America and 7 others occur 
there. 

In the Monocotyledons, then, 57.8 per cent of the 
genera are endemic in South America, another has 8 of its 
10 species there, while the two others may possibly have 
been derived from the same region; 63.1 per cent of the 
apically dehiscent monocotyledonous genera occur in South 
America. Among the Dicotyledons, we have 27.5 per 
cent endemic in South America and 7 more genera have 
some of their species, and in five cases the most of their 
species, in South America. Together, 18 of the 40 apically 
dehiscent genera of Dicotyledons are represented in South 
America, or 45 per cent. 

Taking both Monocotyledons and Dicotyledons, we have 
30 of the 59 genera assigned to this type represented in 
South America, or 50.8 per cent. 

These genera contain from one to over nine hundred 
species; in some, all of the species are apically dehiscent, 
while in others, as in Begonia of about 400, and Ardisia 
of about 200 species, only a few species are apically 
dehiscent. To this question of the geographical distribu- 
tion of species, we shall recur later. Here we may merely 
note that Solatium has about 700 of its 970 species in 



236 MISSOURI BOTANICAL GARDEN. 

South America, while Cassia has about 300 out of 412 iu 
the same region. 

In comparison with the above forms the distribution of 
the Melastomataceae is of great interest. 

The family belongs to the tropical zone but is also repre- 
sented in the subtropical, occurring between 30° N. and 
40° S. Latitude. The family reaches its highest differen- 
tiation in tropical America where 6(1, IV, V, VII, X, XI) 
of the tribes are exclusively found, 1 more (III) occurs 
in South America and extratropical North America, and 
another (XV) is found in tropical America, the Antilles, 
islands of the Pacific Ocean and tropical Africa. The 
other 7 are confined to the Old World. The tropical 
American tribes include 95 genera and three more are 
added by the two tribes of more general distribution while 
the single genus {Rhexia~) in North America brings the 
number of American genera up to 99, with about 2,000 
species all but 7 of which occur south of the United States 
(in this included the species of the Antilles) as com- 
pared with 62 genera with about 700 species from the 
remainder of the range. It is of interest to note that in 
the Memecyleae (Tribe XV), the only South American 
tribe occurring elsewhere, the dehiscence in the South 
American genus ( Olisbea is by some re garded as merely 
a section of 3£ouriria} is by short slits which in many 
cases much resemble pores while in the other genera it is 
described as by longitudinal slits. Of this tribe there are 
44 American and 101 Old World species. It will be noticed 
that in the tribes showing dehiscence by longitudinal 
slits (XII, XIV, XV) 2 genera with 44 species are Amer- 
ican while 10 genera with 156 species are found elsewhere. 

These figures are certainly suggestive, and seemed more 
so when it was found that the Apidae, the insects of the 
greatest importance in the fertilization of these forms, are. 
strongly represented in South America, India and Austra- 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 237 

lia. I resolved to carry out, so far as the nature of the 
data might permit, a statistical investigation of the problem. 

At the present time, the importance of statistical methods 
is being emphasized by many writers. In floral ecology, 
attempts have long been made, and very successfully, too, 
in many cases, to estimate exactly the relative importance 
of each class of visitors in the evolution of a given flora. 
The problem at present in hand differs essentially from 
those which have preceded it in that it is an attempt to 
separate certain floral types and estimate their relative 
numerical importance in any flora. 

The difficulties encountered are very great and the result 
obtained with the present material is not wholly complete. 
The following is the method employed and the results it 
has yielded. 

The first requisite was the determination of the distri- 
bution of the apically dehiscent forms . 

For the sake of convenience, the earth was divided into 
fourteen regions. It hardly need be remarked that in the 
present state of our knowledge any division into great and 
sharply defined phytogeographic regions is purely artificial 
and intended to serve merely as a convenience for further 
study. Fourteen divisions were recognized, since, after 
careful consideration, it seemed best to adopt so far as 
possible the floristic regions of Drude. 

It was deemed best to have the regions represent, so far 
as possible, systematic rather than ecological units. It 
should be remembered that my own data had no influence 
in the selection of the divisions to be used. The principal 
need which led to the employment of a series of regions 
was the desirability of making a statistical comparison of 
the percentages of apically dehiscent genera belonging to 
the various types in the several floras. To facilitate tabu- 
lation, I have modified the regions as outlined by Drude by 
making their 'limits conform to geographical, or, in some 
cases, even political boundaries. Without this precaution, 



238 MISSOURI BOTANICAL GARDEN. 

it would hardly have been possible to gain any numerical 
idea of the relative frequency of the various forms as com- 
pared with the other members of the floras. These modi- 
fications have seemed perfectly justifiable in view of the 
purpose which the regions are to serve. 

The divisions have been lim ited as follows : — 

I. TROPICAL REGIONS. 

1. Tropical African Region. Arabia and the continent of 

Africa south of the Tropic of Cancer to Cape Colony, South 
Bechuanaland, South African Republic, BasutoLand, and Cape 
Colony; also Sokotra Island. 

2. Bast African Island Region, a, Madagascar; b, Mas- 

carine Islands and Rodrigues ; c, Seychelles Islands. 

3. Indian Region. Western and eastern peninsulas of India and 

all islands lying between 17° N. lat. and the Tropic of Capri- 
corn. 

4. Tropical American Region, a, Brazil, Uruguay, Ecuador; 

b, Guiana, "Venezuela, Colombia, Central America; c, Antilles 
and Bahamas; d, Mexico and Southern California. 

II. AUSTRAL REGIONS. 

5. South African Region. Continental South Africa south of 

the tropical region, and the island of St. Helena. 

6. Australian Region. Australia and Tasmania. 

7. Austral American Region. Argentina, Chile, Patagonia and 

Tierra del Fuego. 

8. New Zealand Region. The island. 

9. Antarctic Region. Land masses lying south of the above 

austral regions. 

III. BOREAL REGIONS. 

10. Mediterranean-Oriental Region. The Atlantic Islands (Ca- 

naries to Azores'), Africa and Arabia north of the tropical 
region (Tropic of Cancer), Spain, Italy, Turkey in Europe 
and Asia, Transcaucasia, Persia, Baluchistan, Afghanistan, 
and the lands included by the outer limits of these. 

11. Central Asian Region. Turkestan, Mongolia and Tibet. 

12. East Asian Region. China, Manchuria, Korea, and Japan 

except Sakhalin Island. 

13. Central North American Region. Roughly commensurate 

with the United States. 

14. Northern Region. Land masses of Europe, Asia and America 

above the limits of the four boreal subtropical regions 
described above. 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 239 

It is evidently of great importance to ascertain just what 
per cent of the total number of plants known is to be 
found in each of these regions. The per cent of apically 
dehiscent forms occurring there can then be compared 
with this and it may be readily seen whether it is relatively 
greater or less than in other regions. A difference in the 
number of genera with a given floral habit in two floras 
which are being compared has little significance if the 
relative number of genera in the two floras varies in the 
same ratio. There must be some basis for comparison ; 
the general conditions of life permit of a greater differ- 
entiation in some regions than in others of the same area 
and the extent of this differentiation must be known to 
serve as a basis for comparisons of the development of 
particular elements. 

Being unable to find any suitable statement of the rela- 
tive richness of the floras of different areas it became 
necessary to tabulate them especially for the present pur- 
pose. A comparison by species would probably be the 
most satisfactory, but it soon became evident that this 
presented too many difficulties. I therefore decided to 
tabulate the genera only and to take the data for this pur- 
pose from Die Natiirlichen Pflanzenfamilien. 

In the preparation of these tabulations the 8,541 genera 
of flowering plants were first set down in order followed 
by the numbers of the regions in which they occur. These 
were then summarized by families, the number of genera 
endemic and occurring in each region being indicated. 

I present in Table A a summary of the distribution of all 
of the flowering plants arranged according to regions. In 
the first column is given the number of the several re- 
gions, in the second the number of endemic genera and the 
per cent which this is of the total number of genera of 
flowering plants, in the third column the number of genera 
occurring but not endemic, and in the fourth column the 



240 



MISSOURI BOTANICAL GARDEN. 



total number of genera occurring and the per cent which 
this is of the whole number of plants. 

TABLE A. 

ALL GENERA O! FLOWERING PLANTS. 



Region. 


Endemic. 


Others. 


Total. 


1 


542 = 6.34% 


1116 


1658 = 19.41% 


2 


259 = 3.03% 


579 


838 = 9.81% 


3 


1126 = 13.18% 


1344 


2470 = 28.91% 


4 


1968 = 23.04% 


1160 


3128 = 36.62% 


5 


394 = 4.61% 


468 


862 = 10.09% 


6 


444= 5.19% 


712 


1156 = 13.53% 


7 


175 = 2.04% 


506 


681 = 7.97% 


8 


23 = • .27% 


187 


210 = 2.45% 


9 


4 = .04% 


21 


25 = .29% 


10 


455 = 5.32% 


819 


1274 = 14.91% 


11 


68 = .79% 


549 


617 = 7.22% 


12 


171 = 2.00% 


846 


1017 = 11.90% 


13 


335 = 3.92% 


871 


1206 = 14.12% 


14 


73 = .85% 


654 


727= 8.51% 



The second table, B, gives the number and the per 
cent of the 59 genera of the Solanum-Cassia type endemic 
and occurring in the several regions. 

TABLE B. 

GENERA OF SOLANUM-CASSIA TYPE. DISTRIBUTION. 



Region. 


Endemic. 


Others. 


Total. 


1 


2 = 3.38% 


8 


10 = 16.94% 


2 


1 = 1.69% 


7 


8 = 13.55 % 
16 = 27.1lf 
26 = 44.06;? 


3 


7 = 11.86% 


9 


4 


17 = 28.81% 


9 


5 


2 = 3.38% 


3 


5 = 8.47 /o 


6 


12 = 20.33% 


6 


18 = 30.50% 


7 


4 = 6.77% 


4 


8 = 13.55% 


8 
9 





2 


2= 3.38% 


10 





2 


2= 3.38% 


11 





2 


2 = 3.38% 


12 


1 = 1.69% 


4 


5= 8.47% 


13 





4 


4= 6.77% 


14 


.... 


1 


1 = 1.69% 



THE DEHISCENCE OF ANTHERS BY APICAL, PORES. 



241 



A comparison of the per cents of this table with those 
of the one representing the distribution of all of the flower- 
ing plants is facilitated by subtracting the per cent of all 
genera in a region from the per cent of apically dehiscent 
forms and so indicating the relative abundance of the latter 
in plus and minus quantities, as in Table C. 



TABLE C. 

GENERA OF SOLANUM-CASSIA TYPE. RELATIVE ABUNDANCE. 



Region. 


Endemic. 


Total. 


1 


— 2.96 


— 2.47 


2 


— 1.34 


+ 3.74 


3 


— 1.32 


— 1.80 


4 


+ 5.77 


+ 7.44 


5 


— 1.23 


— 1.62 


6 


+ 13.14 


-f- 16.97 


7 


+ 4.73 


+ 5.58 


8 


— .27 


+ .93 


9 


— .04 


— .29 


10 


— 5.32 


— 11.53 


11 


— .79 


— 3.84 


12 


— .31 


— 3.43 


13 


— 3.92 


— 7.35 


14 


— .85 


— 6.82 



This reveals some suggestive points. In all regions ex- 
cept 4, 6 and 7, the per cent of apically dehiscent genera 
endemic is below that of all forms endemic in the same 
region. In all regions except 2, 4, 6, 7 and 8, the per 
cent of all apically dehiscent genera found is lower than 
that of other forms occurring in the same regions. In 
regions 4, 6 and 7, the average relative abundance of 
endemic apically dehiscent forms is 8.54 more than the 
average for the whole number of endemic forms while in 
the other eleven regions it is 1.66 less. For the apically 
dehiscent genera occurring in regions 4, 6 and 7 the average 
is 9:99 above while for the other eleven regions it is 3.22 
below that of all genera occurring. 

The great specific differentiation of Solanum, Cassia and 

16 



242 MISSOURI BOTANICAL GARDEN. 

of many melastomataceous genera in tropical America first 
called my attention to the problem of the geographical 
distribution of these floral types. 

It would appear that in general the most satisfactory 
statistical comparison of the floras of two or more regions 
is to be made not on the basis of genera but of species. 

While this proposition would seem to hold for general 
comparisons of floras, we have in the present problem a 
special case in which the morphological differentiation of 
Certain organs of the species entering into the composition 
of the floras rather than the taxonomic differentiation of 
the floras as a whole is to be considered. The comparison 
in this case should be based upon morphological rather 
than taxonomic units, but it becomes at once clear that 
it is quite impossible to separate the two conceptions. 

I think that on the whole the special data which we seek 
are furnished by the characteristics of genera rather than by 
the peculiarities of species. In the limitations of the former, 
the floral organization has been assigned a more impor- 
tant place than in the separation of the latter where the 
vegetative features are usually more minutely studied 
and described. But even if all species of a genus have an 
identical floral structure it must not be forgotten that spe- 
cific differentiation may permit of adaptation to more varied 
environmental conditions and so increase the relative 
importance of the genus as a constituent of the flora where 
the area under consideration is so large as to present a 
multiplicity of conditions. 

In a comparison which embraces ecological considerations, 
as the present one does, there is one strong objection to 
taking the number of genera as data. In ecological com- 
parisons, dominance is obviously of much significance. In 
floral ecology where the anthophilous fauna is the environ- 
ment, one species may have characteristics which enable it 
to secure more visits than a dozen others. It is obviously 
incorrect to present merely a numerical statement of the 



THE DEHISCENCE OP ANTHERS BY APICAL POKES. 243 

relative number of genera or even species represented in a 
region as indicative of the ecological conditions prevailing 
there. Only direct field observations of the fauna and 
flora will yield the exact data requisite for a satisfactory 
solution of the problem of the mutual interdependence of 
the distribution of floral structures and the organic envi- 
ronment to which they are adapted. Even these field 
studies cannot lay before us the past history of the species 
or structures in question and our knowledge must always 
lack the exactness of some other phases of biological inves- 
tigation. These limitations notwithstanding, we should 
attempt to bring our knowledge to the greatest exactness 
possible. The broad outlines of a problem are often 
obscured for the naturalist in the field by the details of his 
observations and must be formulated by the worker in the 
herbarium and library. 

I have been unwilling to forego an attempt at a compari- 
son of the several regions by species. The difficulties in 
the way of such an undertaking are known only to those 
who have made a similar attempt. In the very beginning 
of the present study, it became evident that a comparison 
of the regions by per cents of the apically dehiscent species 
in the flora was quite out of the question, and even a com- 
parison by genera, rough and unsatisfactory as it is, has 
required many months more than was originally antici- 
pated. 

The accompanying table, D, is a summary of the number 
of species of the Solanum-Cassia type occurring in each 
region and the per cent of the total number of apically 
dehiscent species of this type known, 1,827. 



244 



MISSOURI BOTANICAL GARDEN. 



TABLE D. 

SPECIES OF THE SOLANUM-CASSIA TYPE. 



Region. 


Species. 


Region. 


Species. 


1 
2 
3 
4 
5 
6 
7 


107 = 5.85% 
42 = 2.29% 

163= 8.92% 

1142 = 62.50% 

37 = 2.02% 

186 = 10.18% 

100 = 5.47% 


8 
9 
10 
11 
12 
13 
14 


2 = .10% 

40 = 2.18% 

4= .21% 

13 = .71% 

29 = 1.58% 

5 = .27% 



This table shows that tropical America has 62.5 per cent 
of all the apically dehiscent species of the Solanum-Cassia 
type, the Indian region 8.9 per cent, the Australian region 
10.1 per cent, the extratropical American region 5.47 
per cent, while the other ten regions average but 1.52 
per cent. 

Solanum and Cassia comprise 1,382 of the 1,827 species 
of the foregoing tables. Anticipating the suggestion that 
the great differentiation of Solanum and Gassia in tropical 
America may account for the very high percentage of 
apically dehiscent species found there, I have given in 
Table E the distribution in numbers and per cents of all the 
apically dehiscent species except those of these two genera 
and, for comparison, have figured the per cents of these 
genera occurring in each region. 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 



245 



TABLE E. 

SPECIES OB" SOLAKFTTM AND CASSIA COMPARED WITH OTHERS OB' SAME 

TYPE. 




The similarity of these per cents is very striking, and, 
as it seems to me, significant. 

For comparison with this I had selected at random from 
the indices of the several phanerogamous volumes of Die 
Naturlichen Pflanzenfamilien, 330 genera, the distribution 
of the species of these tabulated and the average number 
of species per genus in each region obtained. These 
averages are based upon too small a series to be of great 
value for comparison, but an estimate of the number of 
species for each region made on the basis of the number 
of genera occurring and the average number of species per 
genus, as shown by the sample selected, indicates that there 
is not the wide difference in the floras in general that is 
shown by the special forms under consideration. 

The evidence from the Solanum-Cassia type seems clear 
and unmistakable. We may now apply this method of 
comparison to the two other floral types considered in this 
paper. 

To the Dilleniaceous type have been assigned 16 genera 



246 



MISSOURI BOTANICAL GARDEN. 



represented by 376 species, found in ten of the fourteen 
regions as indicated in Table F. 



TABLE E. 

DILLKNIACBOU8 TYPE. 



Region. 


Genera. 


Species. 


1 
2 
3 
4 
5 
6 
7 
8 
11 
12 


4 — 25.00% 
2 = 12.50% 
10 = 62.50% 
6 = 37.50% 
1 = 6.25% 
6 = 37.50% 

1 = 6.25% 

2 = 12.50% 

3 = 18.75% 
3 = 18.75% 


14= 3.72% 

15= 3.99% 
220 = 58.50% 

75 = 19,95% 
3= .77% 

22 = 5.85% 
7 = 1.86% 
6= 1.60% 
8= 2.13% 
6= 1.60% 



It is unnecessary to compare the figures for this type 
with those for all plants. It will be seen that there is con- 
siderable similarity to the results obtained for the Solarium- 
Cassia type, but several points of difference may be in part 
due to the small number of genera and species assigned to 
this type. 

The distribution of the Melastomataceae has already 
been considered in general terms. The distribution of the 
152 genera assigned to the Melastomataceous type is 
shown in Table Gr. 

TABLE G. 

GENERA. OB" MELASTOMATACBQTJS TYPE. 



Eegion. 


Endemic. 


Others. 


Total. 


1 
2 
3 
4 
5 
6 
12 
13 


13= 8.55% 

8 = 5.26% 

22 = 14.47% 

96 = 63.15% 

3 = 1.97% 

• ••• ••■• 

' 1 = !65 % 


6 
2 
9 
1 
1 
4 
6 



19 = 12.50% 

10= 6.67% 

31 = 20.39% 

97 = 63.81% 

4 = 2.63% 

4 = 2.63% 

6= 3.94% 

1 = .65% 



THE DEHISCENCE OP ANTHERS BIT APICAL PORES. 247 

These figures require no comment. The distribution of 
the species of the Melastomataceae, I have not figured. 
Cogniaux in his monograph of the family recognizes about 
2,000 species from tropical America and 700 from Asia, 
Africa, Oceanica and Australia. The per cent of species 
in the fourth region is higher for this type than for any 
other, being roughly 73, while the remaining 27 per cent 
are found almost exclusively in the first, second and third 
regions. 

Summarizing the distribution of the 227 apically dehis- 
cent genera of all three types, we obtain the figures given 
in Table H. 

TABLE H. 

SUMMATION OF GENERA, ALL TYPES. 



Begion. 


Endemic. 


Others. 


Total. 


1 


16 = 7.04% 


17 


33 = 14.53% 


2 


9 = 3.96% 


11 


20 = 8.81% 


3 


34 = 14.97% 


23 


47 = 20.70% 


4 


116 = 51.10% 


13 


129 = 56.82% 


5 


5 = 2.20% 


5 


10 = 4.40% 


6 


13 = 5.72% 


16 


28 = 12.33% 


7 


4 = 1.76% 


5 


9 = 3.96% 


8 

9 

10 





4 


4 = 1.76% 


• • • . •**• 


*2 


V=L" '.88% 


11 


.... .... 


5 


5 = 2.20% 


12 


1 = .44% 


13 


14 = 6.16% 


13 


1 = .44% 


4 


5 = 2.20% 


14 




1 


1 = .44% 



The Geographical Distribution op the Apidae. 

I was early impressed by the great number of Apidae, 
especially the larger species, occurring in South America. 
A preliminary tabulation of the tenth volume of Dalia 
Torre's Catalogus Hymenopterorum confirmed the impres- 
sion derived from a casual examination of various special 
monographs, and the Apidae were then tabulated from 
this work according to the same regions as those used 



248 



MISSOURI BOTANICAL GARDENv 



for the plants. It hardly need be remarked that the 
regions recognized for the Apidae are identical with those 
used for the plants simply to facilitate comparison and not 
from any evidence derived from the classification or mor- 
phology of vthe former. 

I present here in condensed form, in Tables I and J, a sum- 
mary of the tabulations of the Apidae alone, and, by way of 
comparison, of all the genera of Hymenoptera (including* 
Apidae) treated in Dalla Torre's voluminous work. These 
are arranged by regions in the same manner as the tabu- 
lations of the plants considered above. 

To the Apidae belong 137 genera and to all Hymenop- 
tera, including the Apidae, 2,407 genera. 



TABLE I. 

GKNBRA OF APIDAE. 



Region. 


Endemic. 


Others. 


Total. 


1 




25 


25 = 18.24% 


2 


' 3 = 2. 19% 


11 


14 = 10.21% 


3 


... ..... 


23 


23 = 16.78% 


i 


27 = 19.77% 


37 


64 = 46.71% 


5 


... ..... 


17 


17 = 12.40% 


6 


12 = 8.76% 


22 


34 = 24.81% 


7 


15 = 10.95% 


24 


39 = 28.46% 


8 





4 


4= 2.91% 


9 


... ...•• 


• - • 


... . . . . 


10 


4 = 2.91% 


47 


51 = 37.22% 


11 


... 


38 


38 = 27.73% 


12 


... • • . . 


26 


26 = 18.97% 


13 


4= 2.91% 


36 


40 = 29 19% 


14 


4= 2.91% 


46 


50 = 36.49% 



THE DEHISCENCE OP ANTHERS BY APICAL POKES. 



240 



TABLE J. 

ALL 6BKERA OP HYMENOPIERA. 



Region. 


Endemic. 


Others. 


Total. 


1 


44= 1.82% 


173 


217= 9.01% 


2 


7 = .29% 


85 


92 = 3.82% 


3 


120= 4.98% 


268 


388 = 16.12% 


4 


283 = 11.75% 


462 


745 = 30.95% 


5 


10= .41% 


122 


132= 4.51% 


6 


60= 2.07% 


186 


236= 9.80% 


7 


40= 1.66% 


130 


170= 7.06% 


8 

9 

10 


6= .24% 


45 


51 = 2.10% 


125= 5.19% 


383 


508 = 21.10% 


11 


14= .58% 


92 


116 = 4.40% 


12 


29= 1.20% 


130 


159 = 6.60% 


13 


210= 8.72% 


631 


841 = 34.94% 


14 


699 = 29.04% 


729 


1428 = 59.30% 



Expressing the distribution of „ the Apidae as compared 
with that of the Hymenoptera in plus and minus quantities, 
we have the accompanying table, K. 

TABLE K. 

GENERA OF APIDAE. RELATIVE ABUNDANCE. 



Region. 


Endemic. 


Total. 


1 




1.82 


+ 9.23 


2 


+ 


1.90 


+ 6.39 


3 




4.98 


+ .66 


4 


+ 


8.02 


+ 15.76 


5 




.41 


+ 7.89 


6 


+ 


6.69 


+ 15.01 


7 


+ 


9.29 


+ 21.40 


8 




.24 


+ .81 


9 








10 


— 


2.28 


+ 16.12 


11 


— 


.58 


+ 23.33 


12 


— 


1.20 


+ 12.37 


13 


— 


5.81 


— 5.75 


14 


~~ ~~ 


26.03 


— 22.81 



250 MISSOURI BOTANICAL GARDEN. 

Regions 3, 4, 6 and 7 average + 4.75 for endemic genera, 
while the remaining nine regions from which Hymenoptera 
have been recorded average — 4.06. Turning to the total 
number of forms occurring in the several regions we find 
that 3, 4, 6 and 7 average + 13.17, while the other nine 
regions average + 5.28. 

Conclusions concerning the distribution of the Hymen- 
optera must be only tentative since the vast order has been 
very little investigated, even in its most highly specialized 
family, the Apidae. The distribution by species need not 
be discussed in this place. 

The points of agreement of these several tables are too 
apparent to need further comment. To facilitate com- 
parison, the distribution of plants and insects has been 
summarized in the form of percentage curves. A glance 
at the accompanying diagrams, L and M, will make clear 
the relative proportion of the different elements in the 
flora and fauna and show the striking similarity between 
the distribution of the apically dehiscent genera and the 
Apidae. In this series of curves, the fourteen ordinates 
represent the fourteen regions, and the percentage of 
plants or insects endemic or occurring in a given region is 
indicated on these by the curves . A is the curve for the 
apically dehiscent forms, B for all genera ; a for the 
Apidae, and b for all Hymenoptera. The high percentage 
of Apidae and other Hymenoptera in regions 10 to 14 is 
doubtless to be accounted for by the greatei' thoroughness 
with which these regions have been studied. 

Comparison of Distribution of Fauna and Flora. 

We seem to have in the data which have just been pre- 
sented strong indications of a direct relation between the 
geographical distribution of the Apidae and that of the 
floral types which we have studied. 

It is somewhat surprising that so little has been done 
towards the investigation of the distribution of floral 



diagram l,. 




Distribution of Endemic Genera. 



DIAGRAM M. 




Distribution of Ali, Genera. 



THE DEHISCENCE OF ANTHERS BT APICAL PORES. 251 

structures. The relation of flowers to insects is one of the 
oldest phases of ecology, and while Sprengel's beautiful 
observations of over a century ago were neglected until the 
theories and researches of Darwin and his contemporaries 
showed their importance, it was not long after the revival 
of interest in this fascinating branch of natural history 
that students began to see that if flowers show a reduced 
degree of fertility with their own pollen and are so closely 
adapted to some special visitor or group of visitors that 
others are unable to pollinate them, their geographical 
range will depend largely upon the distribution of the ani- 
mals which are able to effect their pollination. That this 
phase of morphology and geography should have been 
neglected seems all the more remarkable in view of the 
significance assigned to floral structures in classification. 

In the sixth chapter of his Plant Geography, Schimper 
considers animals as factors in plant distribution. Certain 
floral types, he considers, are undoubtedly limited in their 
geographical distribution by the range of the animals 
which pollinate them. The problem of adaptation for the 
protection of the plant from animal enemies has been much 
less satisfactorily investigated except in the case of 
myrmecophilism. 

One of the best classes of data for the investigation of 
this problem is that furnished by ornithophilous flowers, 
since birds which are of importance as pollinators are of 
restricted range, and the adaptations, when they are real 
and not merely apparent, are evidently dependent in their 
range upon that of the organisms to which they are adapted. 
Comparatively few plants are ornithophilous, but the evi- 
dence offered is of such a character as to give us confi- 
dence in attacking the more diflicult problem offered by 
entomophilous forms. Entomophily is the more general 
condition and in its geographical range is practically coex- 
tensive with the limits of the Phanerogams. Only three 
orders of insects seem to have influenced essentially the 



252 MISSOURI BOTANICAL GARDEN. 

structure of flowers, the Diptera, Lepidoptera and Hyrnen- 
optera. These groups are of general distribution, and 
species with flowers adapted to representatives of each of 
them may be found in almost every flora. The problem 
here is much more involved. Conclusions must, in this 
case, rest upon the demonstration of the interdependence 
of fauna and flora and statistical comparisons of the floras 
and faunas of different regions. 

It is well known that the three orders of insects men- 
tioned are represented in very different proportions in 
different regions, and the investigations of several workers 
have shown that there are corresponding differences in the 
constitution of the flora. 

It is Hermann Miiller to whom our knowledge of the 
influence of the insect fauna upon the constitution of the 
flora is largely due, for it was his contribution of the statis- 
tical methods of work which has made possible much which 
has been done since his time. So long as observations 
were confined to individual species showing adaptations to 
a single insect visitor or groups of visitors, conclusions 
were necessarily much limited and unsatisfactory. The 
development of methods which would permit of a statis- 
tical comparison of the adaptations of different flowers as 
a whole to insect visitors is clearly a step of fundamental 
importance, and, if the methods be trustworthy, permits 
of advances and conclusions of an accuracy otherwise quite 
unattainable, though the number of factors is so great and 
conditions are so complex that comparisons must remain 
only approximate. 

In several papers, Miiller developed various phases of the 
statistical methods, but it was in his book on Alpineflowers 
that he especially applied and illustrated these methods, and 
a review of the theoretical portion of this classic work will 
explain and illustrate the statistical method as developed 
by the German naturalist. Thomson, Loew, Warming, 
Landman, Bonnier, MacLeod, Willis and Burkhill, Knuth, 



THE DEHISCENCE OF ANTHERS BY APICAL POKES. 253 

Robertson and others have made valuable contributions to 
this branch of statistical investigation. 

The scope of this paper does not include a general re- 
view of the several attempts which have been made to 
demonstrate a relationship between the geographical dis- 
tribution of flowers and the insect environment. MacLeod 
in the introduction to his careful investigation of the in- 
sect relations of the flowers of the Pyrenees has given such 
a review of the more important papers. Schimper and 
Drude have doubtless performed a great service for this 
phase of biological investigation by recognizing animals as 
representing one of the factors in physiological plant 
geography. 

In this paper I have sought to approach the problem 
from a point of view different from that usually taken. 
Previous researches have been almost exclusively exami- 
nations of the range of individual species or genera 
which are clearly adapted exclusively to a single polli- 
nating agent, with the range of the organism to which they 
are adapted, or statistical comparisons by the methods 
of Miiller and MacLeod of floras of different regions, 
lowland and alpine, insular and continental, temperate and 
tropical and boreal. 

I have tried to select all floral forms of a given type and 
after assuring myself of the reality of their morphological 
semblance and, so far as our limited knowledge of the 
plants in their living state would permit, of the similarity 
of their ecological relationships, I have considered their 
geographical distribution throughout the main divisions of 
the globe. 

This method as applied to the problem of the mor- 
phology and biology of the apically dehiscent anther has 
yielded results which are, I think, of some importance in 
the larger problem of the evolution of floral structures. 

Any problem in geographical distribution is necessarily 
dependent upon the validity of taxonomic or morphological 



254 MISSOURI BOTANICAL GARDEN. 

data, and writers who may speak with the greatest authority 
have repeatedly insisted that biogeographical researches are 
of the greatest value only when prosecuted by those who 
are deeply versed in the morphology and taxonomy of the 
forms treated. When the problem is one of physiological 
plant geography, and we have for consideration the addi- 
tional factor of the environmental conditions and adapta- 
tions thereto, the problem becomes more complex. Upon 
the morphological and ecological data of the problem I 
have hardly touched in the present paper. I have sought, 
however, to carry out my comparisons with the greatest 
care, and I feel confident that the publication of the whole 
mass of evidence with all the exceptions to the more gen- 
eral laws and transgressions of the categories recognized 
would convince others of the validity of the propositions 
offered in this paper. 

A point which will doubtless occur to all who examine 
into this phase of biology is that in the present work I 
have compared the morphology and geographical distribu- 
tion of a single type of adaptation to a given factor, — 
believing that the Solanum-Cassia and Melastomataceous 
types may be considered as adapted to the peculiar habits 
of the Apidae, — with the form and distribution of all floral 
types, whether representing adaptations or not. For a 
final solution of the problem this method is obviously in- 
adequate. It seems at first that the comparison should be 
confined to the entomophilous forms and that these should 
be segregated into their various types of adaptation for 
comparison; but the relation of all floral structures is one 
of such complexity that it seems that even this limitation 
of the field may be too narrowly drawn. Concurrence in 
both fauna and flora is here doubtless a factor of great im- 
portance, and far-reaching structural and field investigations 
must be made before the real relations will be finally made 
clear. In justification of the course which has been fol- 



THE DEHISCENCE OP ANTHERS BY APICAL PORES. 255 

lowed in this study — if indeed any attempt at arrange- 
ment and interpretation of one class of data preliminary 
to the elaboration of all the material needs justification — 
it may he pointed out that these apically dehiscent types 
represent only two or three of the several types of flowers 
adapted to Apidae. Hypothetically all these types are 
well represented in regions which are rich in Apidae, and 
the relative values obtained for any type or types, as the 
Solanum-Cassia or Melastoinataceous, would be too small 
rather than too large as an indication of the total influence 
of this factor of the environment, since the several types of 
adaptation to this factor, for instance the Solanum-Cassia 
and the Verbascum type among pollen flowers adapted to 
Apidae, will be in direct competition with each other as 
well as with all other types of adaptation. 

Space has not permitted of the discussion of points of 
theory, such as the nature of the correlation between the 
different floral parts, the amount and nature of the 
influence of the fauna upon the evolution of these floral 
types, or the significance of these conclusions for current 
conceptions of floral evolution, nor would such be desir- 
able in the absence of a more detailed presentation of the 
data upon which the conclusions actually drawn are based. 
In my final paper, I hope to point out the nature of some 
of these problems and to offer some suggestions as to the 
interpretation of the evidence already accumulated. 

Summary. 

The three objects of the present investigation seem to be 
satisfactorily accomplished, and there only remains the 
working out of the large number of minor and detailed 
structural, ecological and geographical problems. A 
descriptive bibliographic list of apically dehiscent forms is 
nearly completed and the available material for a mono- 
graphic treatment of their floral ecology brought together. 



256 MISSOURI BOTANICAL GARDEN. 

The work on the gross structure is being supplemented by 
a histological investigation of the anthers of the forms in 
question, and some hundreds of microscopical slides have 
been prepared and examined. Because of the great variety 
and in many cases complexity of structure this promises 
to be one of the most important phases of the work. 

It has been shown that apically dehiscent forms for the 
most part fall into well defined categories exhibiting a 
striking uniformity of structure in genera or species from 
widely separated families. Structural parallelism in a series 
of floral forms assembled from phylogenetically widely sep- 
arated groups necessarily implies the anomalous structure 
of many of the forms in the groups to which they syste- 
matically belong, among the forms to which they are genet- 
ically related. Of the real process hj which this paral- 
lelism has been evolved I do not yet care to say anything. 
Many of the conclusions drawn from ecological studies are 
of little value because of the inadequacy of the data con- 
sidered. Comparative researches attain their fullest value 
only when based on the largest series of material. At the 
present time, it seems to me, one of the most important 
things is to determine in how far floral organizations do 
really fall into well defined groups and in how far the 
structural characteristics of such groups may be regarded 
as adaptations. The first of these problems may be an 
herbarium and librai-y investigation while the solution of 
the second is to be approached only by field studies. 

Of the third and last problem little need be said. The 
tables presented show clearly that the apically dehiscent 
genera and species are not only much more abundant 
numerically in tropical America, Australia and the Indian 
region, but also that they form there a larger proportional 
constituent of the flora than they do in other regions. This 
distribution seems to depend in large measure upon the 
relative abundance of the Apidae in the several divisions. 



THE DEHISCENCE OF ANTHERS BY APICAL PORES. 257 

So far as a biological science can be an exact science, 
plant geography and ecology must become exact sciences. 
One of the most promising methods of investigation, it 
seems to me, is to divide plant structure into classes 
limited as sharply and as naturally as possible, and to study 
the ecology and geographical distribution of these. When 
this is not done, the confusion of details is apt to obscure 
the results of the greatest value. 

17